# Обобщение основной теоремы арифметики Буланкина Вера, Зайцев Тимофей, Фролов Иван,
Петухов Алексей, Салимов Руслан* ## Введение Цель этого проекта - обобщить основную теорему арифметики с целых чисел на какие-то более "продвинутые" объекты. Для решения задач проекта можно (пожалуй, даже рекомендовано) объединяться в команды с другими участниками. Основная теорема арифметики (Теорема 1 ниже, см. также её обобщения - Теоремы 4 , 5,7 ) полезна при решении различных уравнений в целых числах. Как мы увидим уже в первой части - примерами могут служить Рождественская теорема Ферма (Теорема 2) и Великая теорема Ферма (Теорема 3) для $n=3$. Вы можете попробовать доказать их, не читая дальнейшего текста, но не тратьте на это много времени. Возвращайтесь к ним по мере чтения проекта. Также мы увидим, что однозначность разложения на простые множители не обязательно сохраняется для рассматриваемых нами аналогов целых чисел. Во второй части проекта будет обсуждена версия основной теоремы арифметики, работающая для чисел вида $a+b \sqrt{d}$, где $d$ - фиксированное целое число, $a, b$ - любые целые (для этого нам потребуется понятие идеала). Эти знания мы применим к решению уравнений в целых числах. В третьей части мы сформулируем более общее утверждение для произвольных алгебраических чисел. Завершающая часть проекта будет посвящена связи общей теории с Великой теоремой Ферма. Вам предлагается доказать так называемый первый случай Великой теоремы Ферма для регулярных простых чисел. Используя похожие идеи можно доказать, что теорему Ферма и во втором случае для всех $n$, делящихся на регулярные простые числа $p$; все простые числа, меньшие 37 - регулярны, см. книги Дж. Милна и М. Постникова. Общее доказательство, придуманное Эндрю Вайлсом, использует существенно другие методы и идеи. При подготовке проекта мы использовали книги К. Айэрланда и М. Роузена, Дж. Милна, М. Постникова, заметки К. Конрада и википедию. Мы также добавили несколько ссылок [SS, Go, ZSS], которые могут оказаться интересны читателю, желающему лучше познакомиться с этой темой. Теорема 1. Основная теорема арифметики. Каждое натуральное число $n>1$ можно представить в виде $n=p_{1} \cdot \ldots \cdot p_{k}$, где $p_{1}, \ldots, p_{k}$ - простые числа, причём такое представление единственно с точностью до порядка следования сомножителей. Теорема 2. Рождественская теорема Ферма. Натуральное число $n$ представляется в виде суммы двух квадратов тогда и только тогда, когда все его простые делители вида $4 k+3$ входят в $n$ в четной степени. Теорема 3. Великая теорема Ферма. Уравнение $x^{n}+y^{n}=z^{n}$ не имеет решений в натуральных числах при $n>2$.[^0] ## 1 Гауссовы числа Мы хотели обсудить в этой части проекта гауссовы числа - обобщение целых чисел, использующее $\sqrt{-1}$. Если у Вас не получится решить какие-то из задач про гауссовы числа, то попробуйте порешать задачи из других частей проекта и вернуться сюда позднее. Определение. Гауссовыми числами называется множество комплексных чисел вида $a+b i$, где $a, b \in \mathbb{Z}, i=\sqrt{-1}$. Будем обозначать множество гауссовых чисел $\mathbb{Z}[i]$. Задача 1. Докажите, что сумма и произведение любых двух гауссовых чисел - гауссово число. Для решения следующей задачи потребуется дать более точную формулировку основной теоремы арифметики. Ведь, строго говоря, в старой формулировке она неверна уже для целых чисел: $2 \cdot 3=6=(-2) \cdot(-3)$. Чтобы получить новую формулировку, нам потребуется новое понятие. Определение. Делители единицы в $\mathbb{Z}[i]$ - такие гауссовы числа $a$, что существует гауссово число $b$ такое, что $a b=1$. Аналогично определяются делители единицы в $\mathbb{Z}$. Упражнение 1. Докажите, что в целых числах делители единицы это 1 и -1 , а в гауссовых добавляются еще $i$ и $-i$. Определение. Гауссово число называется простым, если в любом его разбиении на 2 множителя ровно один является делителем единицы. Определение. Два разложения на простые множители называются одинаковыми, если в них одинаковое число множителей, и их можно так переставить, чтобы отношение соответствующих простых множителей было делителем единицы. Например, $7 \cdot 3,(-3) \cdot(-7)$ и $(-7 i) \cdot(3 i)$ - это одинаковые разложения числа 21 в гауссовых числах. Наша ближайшая цель доказать и научиться применять идущую ниже теорему. Теорема 4. Основная теорема арифметики для гауссовых чисел. Любые два разложения гауссова числа на простые множители одинаковы. Задача 2* Определите деление с остатком для $\mathbb{Z}[i]$. Используя это, докажите Теорему 4. Подсказка: используйте модуль комплексного числа и графическую (гауссову) интерпретацию комплесных чисел. Задача 3. Решите в целых числах уравнение $x^{2}+1=y^{n}$. Задача 4. а) Пусть $p \in \mathbb{Z}$ - простое число. Докажите, что число $p$ является простым гауссовым числом тогда и только тогда, когда $p+1$ делится на 4 . б) Если $n, m \in \mathbb{Z}$ представляются в виде $a^{2}+b^{2}$, где $a$ и $b \in \mathbb{Z}$, то $m n$ представляется в виде суммы двух квадратов. в) Докажите Рождественскую теорему Ферма (Теорему 2). Задача 5. Как по разложению целого числа на (гауссовы) простые множители понять, сколькими способами оно раскладывается в сумму двух квадратов? ## Числа Эйзенштейна В этой главе мы постараемся помочь участникам решить следующую задачу. Задача 6* Докажите Теорему 3 для $n=3$, используя формулу $$ x^{3}+y^{3}=(x+y)\left(x+\frac{-1+\sqrt{-3}}{2} y\right)\left(x+\frac{-1-\sqrt{-3}}{2} y\right) $$ Для этого мы введём несколько новых определений и рассмотрим несколько вспомогательных задач. Обозначим через $\xi$ какой-то комплексный корень третьей степени из 1 , не равный 1. Упражнение 2. Докажите, что $\xi=\frac{-1 \pm \sqrt{-3}}{2}$. Положим $\mathbb{Z}[\xi]:=\{a+b \xi: a, b \in \mathbb{Z}\}$. Определение. Число $a \in \mathbb{Z}[\xi]$ делится на $b \in \mathbb{Z}[\xi]$ тогда и только тогда, когда существует такое $c \in \mathbb{Z}[\xi]$, что $a=b c$. Делители единицы в $\mathbb{Z}[\xi]$ определяются аналогично делителям единицы в $\mathbb{Z}$ и $\mathbb{Z}[i]$. Определение. Число $\alpha \in \mathbb{Z}[\xi]$ составное, если $\alpha=\beta \gamma$, где $\beta$ и $\gamma \in \mathbb{Z}[\xi]$ не являются делителями единицы. Число $\alpha \in \mathbb{Z}[\xi]$ называется простым, если $\alpha$ не составное и не делитель единицы. Задача 7. Определите деление с остатком для $\mathbb{Z}[\xi]$. Используя его, сформулируйте и докажите, основную теорему арифметики для $\mathbb{Z}[\xi]$. Задача 8. Найдите все делители единицы в $\mathbb{Z}[\xi]$. ## Квадратичные расширения В этой секции мы введём общие квадратичные расширения, как обобщения гауссовых чисел и чисел Эйзенштейна. Определение. Для произвольного набора комплексных чисел $a_{1}, \ldots, a_{n}$ обозначим через $\mathbb{Z}\left[a_{1}, \ldots, a_{n}\right]$ множество всех комплексных чисел, получаемых из целых чисел $(\mathbb{Z}$ ), а также $a_{1}, \ldots, a_{n}$ сложением, вычитанием и умножением. Аналогично определяется $\mathbb{Q}\left[a_{1}, \ldots, a_{n}\right]$. Мы рассматриваем такие множества $\mathbb{Z}\left[a_{1}, \ldots, a_{n}\right]$ как аналоги целых чисел (их элементы можно складывать, умножать и вычитать). Фиксируем целое число $d \neq 1$, не делящееся на квадраты простых чисел. Комментарий. Два важных примера: $d=-3$ и $d=2$. Может оказаться полезным продумать и проанализировать все определения этой секции сначала для этих двух примеров, а потом для общего случая. Упражнение 3. а) Докажите, что $$ \mathbb{Q}[\sqrt{d}]=\{x+y \sqrt{d}: x, y \in \mathbb{Q}\} \text { и } \mathbb{Z}[\sqrt{d}]=\{x+y \sqrt{d}: x, y \in \mathbb{Z}\} $$ б) Докажите, что если $a, b \in \mathbb{Q}[\sqrt{d}]$ и $b \neq 0$, то $\frac{a}{b} \in \mathbb{Q}[\sqrt{d}]$. Определение. Целыми числами в $\mathbb{Q}[\sqrt{d}]$ называются $\alpha \in \mathbb{Q}[\sqrt{d}]$, которые являются корнями уравнений вида $x^{2}+p x+q$, где $p, q \in \mathbb{Z}$. Упражнение 4. Является ли $\xi$ целым в $\mathbb{Q}[\sqrt{-3}]$ ? Является ли $\frac{1+i}{2}$ целым в $\mathbb{Q}[i]$ ? Положим $\omega=\sqrt{d}$ если $d \equiv 2,3 \bmod 4$, и $\omega=\frac{\sqrt{d}+1}{2}$ если $d \equiv 1 \bmod 4$. Для $\mathbb{Z}[\sqrt{d}]$ единственность разложения на простые множители не всегда имеет место, но, как мы надеемся, вы сможете доказать некоторую её модификацию. Для каждого числа $\alpha=a+b \sqrt{d}$, где $a, b \in \mathbb{Q}$, определим сопряженное число $\bar{\alpha}=a-b \sqrt{d}$, норму $N(\alpha)=\alpha \bar{\alpha}$ и след $\operatorname{Tr}(\alpha)=\alpha+\bar{\alpha}$. Упражнение 5. Докажите, что $\overline{a+b}=\bar{a}+\bar{b}$ и $\overline{a b}=\bar{a} \bar{b}$. Упражнение 6. а) Докажите, что $\alpha \in \mathbb{Q}[\sqrt{d}]$ - корень многочлена $x^{2}-\operatorname{Tr}(\alpha) x+N(\alpha)$. б) Докажите что $\alpha \in \mathbb{Q}[\sqrt{d}]$ цело тогда и только тогда когда $N(\alpha) \in \mathbb{Z}$ и $\operatorname{Tr}(\alpha) \in \mathbb{Z}$. Задача 9. Докажите, что целые числа в $\mathbb{Q}[\sqrt{d}]$ совпадают с $\mathbb{Z}[\omega]$. Делители единицы в $\mathbb{Z}[\xi]$ определяются аналогично делителям единицы в $\mathbb{Z}$ и $\mathbb{Z}[i]$. Упражнение 7 . Докажите, что для $\gamma \in \mathbb{Z}[\omega]$ выполнено $N(\gamma)= \pm 1$ тогда и только тогда, когда $\gamma$ - делитель единицы. Определение. Число $a \in \mathbb{Z}[\omega]$ делится на $b \in \mathbb{Z}[\omega]$ тогда и только тогда, когда существует такое $c \in \mathbb{Z}[\omega]$, что $a=b c$. Определение. Число $\alpha \in \mathbb{Z}[\omega]$ составное, если $\alpha=\beta \gamma$, где $\beta$ и $\gamma \in \mathbb{Z}[\omega]$ не являются делителями единицы. Число $\alpha \in \mathbb{Z}[\omega]$ называется простым, если $\alpha$ не составное и не делитель единицы. Упражнение 8. Докажите, что если для числа $\gamma \in \mathbb{Z}[\omega]$ число $|N(\gamma)| \in \mathbb{Z}$ просто, то $\gamma$ просто. Докажите что обратное утверждение неверно в $\mathbb{Z}[\sqrt{3}]$. Упражнение 9. Докажите, что если $\gamma \in \mathbb{Z}[\omega] \backslash 0$ не делитель единицы, то $\gamma$ равно произведению каких-то простых элементов $\mathbb{Z}[\omega]$. Задача 10. Проверьте, что все множители в разложении $$ 15=3 \cdot 5=(1+\sqrt{-14})(1-\sqrt{-14}) $$ простые. Задача 11. Определите деление с остатком для а) $\mathbb{Z}[\sqrt{-2}]$; б) $\mathbb{Z}\left[\frac{1+\sqrt{-7}}{2}\right]$. Используя его, сформулируйте и докажите основную теорему арифметики для а) и б). Задача 12. Найдите все делители единицы в а) $\mathbb{Z}[\sqrt{-1}]$, б) $\mathbb{Z}[\sqrt{-d}]$, где $d \geq 1$, в) $\left.\mathbb{Z}\left[\frac{1+\sqrt{-3}}{2}\right], \Gamma^{*}\right) \mathbb{Z}[\sqrt{2}]$. Задача 13. Выполнена ли основная теорема арифметики для а) $\mathbb{Z}[\sqrt{2}]$, б) $\mathbb{Z}[\sqrt{-3}]$, в) $\mathbb{Z}[\sqrt{3}]$, г) $\mathbb{Z}[\sqrt{-5}]$, д) $\mathbb{Z}[\sqrt{5}]$, е) $\mathbb{Z}[\sqrt{10}]$, ж) $\mathbb{Z}\left[\frac{1+\sqrt{5}}{2}\right]$, $$ \text { з) } \left.\mathbb{Z}\left[\frac{1+\sqrt{-7}}{2}\right], \text { и) } \mathbb{Z}\left[\frac{1+\sqrt{-11}}{2}\right], \kappa^{*}\right) \mathbb{Z}\left[\frac{1+\sqrt{-19}}{2}\right] \text { ? } $$ Задача 14. При каких комплексных $\xi$ сумма и произведение любых двух чисел вида $a+b \xi$, где $a$ и $b \in \mathbb{Z}$, снова имеет такой же вид? Задача 15* Решите в целых числах уравнения a) $3^{n}=k^{2}+2$ б) $2^{n}=k^{2}+7$. ## Идеалы В этой и последующих главах намечен подход к решению следующих трудных задач. Задача 16. Решите в целых числах уравнения a) $x^{2}+5=y^{3}$, б) $x^{2}+2 x+7=y^{3}$ в) $5 x^{2}+1=y^{3}$, г) $6 x^{2}-12 x+7=y^{3}$, д) $x^{2}-6=y^{3}$. Определение. Непустое подмножество I множества целых чисел называется идеалом, если оно замкнуто относительно сложения, вычитания и умножения на целые числа: $$ a, b \in I \Longrightarrow a \pm b \in I, \quad a \in I, b \in \mathbb{Z} \Longrightarrow a b \in I $$ Познавательная минутка. Идеалы были придуманы Юлиусом Дедекиндом как формализация "идеальных чисел", придуманных Э. Куммером, как считается, при размышлениях о Великой теоремы Ферма: можно говорить о том, делится или не делится данное число на "идеальное число" вне зависимости от того каков статус этого числа. Аналогичная идея стоит за определением дедекиндова сечения рациональных чисел. Упражнение 10. а) Докажите, что всякий идеал содержит 0. б) Докажите, что если $a \in I$, то $-a \in I$. в) Докажите, что множество чётных чисел - идеал. г) Докажите, что множество чисел вида $2018 m, m \in \mathbb{Z}$, - идеал. Упражнение 11. Докажите, что пересечение идеалов - идеал. Здесь и далее мы обозначаем через $\left(a_{1}, \ldots, a_{n}\right)$ наименьший идеал, содержаций числа $a_{1}, \ldots, a_{n} \in \mathbb{Z}$ (т.е. пересечение всех идеалов, содержащих числа $a_{1}, \ldots, a_{n} \in \mathbb{Z}$ ). Задача 17. а) Пусть $a, b$ - это два взаимнопростых числа. Докажите, что $(a, b)=(1)=\mathbb{Z}$. б) Докажите, что $(a, b)=(d)$, где $a, b \in \mathbb{Z}, d$ есть наибольший общий делитель $a, b$. в) Докажите, что любой идеал $I$ в $\mathbb{Z}$ совпадает с ( $d$ ) для какого-то числа $d \in \mathbb{Z}$. Как в предудыщем разделе, считаем, что $\omega=\sqrt{d}$ если $d \equiv 2,3 \bmod 4$, и $\omega=\frac{\sqrt{d}+1}{2}$ если $d \equiv 1 \bmod 4$. Идеал в $\mathbb{Z}[\omega]$ определяется абсолютно точно так же, как и идеал в $\mathbb{Z}(\mathbb{Z}$ заменяется на $\mathbb{Z}[\omega])$. Точно так же всякий набор $a_{1}, \ldots, a_{n} \in \mathbb{Z}[\omega]$ определяет идеал $\left(a_{1}, \ldots, a_{n}\right)$ в $\mathbb{Z}[\omega]$. В частности, любой элемент $a \in \mathbb{Z}[\omega]$ определяет идеал ( $a$ ). Упражнение 12. Докажите, что $(\alpha)=(\beta)$ тогда и только когда $\alpha / \beta \in \mathbb{Z}[\omega]$ и $\beta / \alpha \in \mathbb{Z}[\omega]$ (т.е. $\alpha / \beta$ - делитель единицы в $\mathbb{Z}[\omega]$ ). Упражнение 13. Пусть $a, x, y \in \mathbb{Z}$. Докажите, что $x+y \omega \in(a)$ тогда и только тогда когда $x$ и $y$ делятся на $a$. Определение. Идеал, имеющий вид ( $a$ ) для какого-то $a \in \mathbb{Z}[\omega]$, называется главным. Как мы видели в Задаче 17 , все идеалы в $\mathbb{Z}$ главные. Задача 18. Докажите, что идеал $(2, \sqrt{-14})$ не является главным в $\mathbb{Z}[\sqrt{-14}]$. Задача 19. Докажите, что для всякого идеала $I$ в $\mathbb{Z}[\omega]$ существуют $\alpha, \beta \in \mathbb{Z}[\omega]$ такие, что $$ I=\{x \alpha+y \beta: \quad x, y \in \mathbb{Z}\} $$ Определение. Для двух идеалов $I, J \in \mathbb{Z}[\omega]$ положим $$ \begin{gathered} I+J:=\left\{i_{1}+i_{2}: i_{1} \in I_{1}, i_{2} \in J\right\}, \quad \bar{I}:=\{\bar{i}: i \in I\} \\ I J:=\left\{i_{1} j_{1}+\ldots+i_{k} j_{k}: i_{1}, \ldots, i_{k} \in I, j_{1}, \ldots, j_{k} \in J\right\} \end{gathered} $$ Упражнение 14. Посчитайте в $\mathbb{Z}[\sqrt{-14}]$ произведение идеалов $I=(5+\sqrt{-14}, 2+\sqrt{-14})$ и $J=(4+\sqrt{-14}, 2-\sqrt{-14})$. Упражнение 15. Проверьте равенства $$ (3)=p_{1} p_{2}, \quad(5)=p_{3} p_{4}, \quad(1+\sqrt{-14})=p_{1} p_{3}, \quad(1-\sqrt{-14})=p_{2} p_{4} $$ где $$ p_{1}=(3,1+\sqrt{-14}), p_{2}=(3,1-\sqrt{-14}), p_{3}=(5,1+\sqrt{-14}), p_{4}=(5,1-\sqrt{-14}) $$ Упражнение 16. Опишите (20)(18), (20) + (18), (20) $\cap(18)$ в $\mathbb{Z}[\omega]$ для всех допустимых значений $d(d \neq 1, d$ не делится на квадраты простых чисел). Упражнение 17. Докажите, что $I+J, \bar{I}, I J$ являются идеалами в $\mathbb{Z}[\omega]$ для любых идеалов $I, J \subset \mathbb{Z}[\omega]$. ## 2 Основная теорема арифметики: квадратичный случай Все рассматриваемые в этой главе идеалы, являются идеалами в $\mathbb{Z}[\omega]$ для подходящего $d$. В Задаче 10 мы убедились, что однозначность разложения на множители в самом очевидном смысле теряется для $\mathbb{Z}[\sqrt{-14}]$ $$ 3 \cdot 5=(1+\sqrt{-14})(1-\sqrt{-14}) $$ С другой стороны, по Упражнению 15: $$ (3)=p_{1} p_{2}, \quad(5)=p_{3} p_{4}, \quad(1+\sqrt{-14})=p_{1} p_{3}, \quad(1-\sqrt{-14})=p_{2} p_{4} $$ где $$ p_{1}=(3,1+\sqrt{-14}), p_{2}=(3,1-\sqrt{-14}), p_{3}=(5,1+\sqrt{-14}), p_{4}=(5,1-\sqrt{-14}) $$ Т.е. $(15)=p_{1} p_{2} p_{3} p_{4}$. Идеалы $p_{1}, p_{2}, p_{3}, p_{4}$ играют роль простых сомножителей, см. Задачу 23 , а разложение ( 15$)=p_{1} p_{2} p_{3} p_{4}$ единственно с точностью до перестановки множителей в соответствии со следующей теоремой. Теорема 5. Основная теорема арифметики для квадратичных расширений. Для каждого идеала $I \subset \mathbb{Z}[\omega]$ существует единственное с точностью до перестановки множителей разложение в произведение простых идеалов $$ I=p_{1}, \ldots, p_{s} \subset \mathbb{Z}[\omega] $$ (Определение простого идеала идёт ниже, Определение 2.) В Теореме 5 есть две существенные части: существование такого разложения и его единственность. Первая доказывается в Задаче 24 , вторая - в Задаче 27 . В качестве примера задачи, которые можно решить, используя эту теорему, мы предлагаем Задачу 16 , см. также завершающий листок про теорему Ферма. Следствие. Для каждого $m \in \mathbb{Z}[\omega]$ существует единственное с точностью до перестановки множителей разложение идеала ( $m$ ) в произведение простых идеалов $p_{1}, \ldots, p_{s} \subset \mathbb{Z}[\omega]$. Доказательство можно разбить на цепочку утверждений, каждое из которых условно просто. В общем и целом схема доказательства похожа на схему доказательства основной теоремы арифметики для целых чисел. Задача 20. Для всякого набора $a_{1}, \ldots, a_{n} \in \mathbb{Z}[\omega]$ докажите, что а) $\left(a_{1}, \ldots, a_{n}\right)\left(\bar{a}_{1}, \ldots, \bar{a}_{n}\right)=\left(N\left(a_{i}\right), \operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)\right)_{1 \leq i, j \leq n}$. Подсказка: разберите сначала случай идеала, порождённого 2 элементами. Определение. Будем говорить что идеал $I$ делится на идеал $J$, если $I=J H$, где $H-$ какой-то идеал в $\mathbb{Z}[\omega]$. Задача 21. Для любых двух идеалов $I, J$ в $\mathbb{Z}[\omega]$, докажите, что $I$ делится на $J$ тогда и только тогда, когда $I$ содержится в $J$. Подсказка: воспользуйтесь предыдущей задачей. Упражнение 18. Используя Задачу 20 докажите, что, для всякого идеала $I$ в $\mathbb{Z}[\omega]$, существует неотрицательное целое число $N(I)$, такое что $I \bar{I}=(N(I))$. Задача 22. Докажите, что идеал $H$ делит $I$ и $J$ тогда и только тогда, когда он делит $I+J$. Упражнение 19. Докажите что $N((a))=|N(a)|$ для всех $a \in \mathbb{Z}[\omega]$. Упражнение 20. Докажите что $N(I) N(J)=N(I J)$ для любых двух идеалов $I, J$. Упражнение 21. Докажите, что если идеал $I$ делит идеал $J$, то $N(I)$ делит $N(J)$. Верно ли обратное? Упражнение 22. Докажите, что $N(I)=1$ тогда и только тогда, когда $I=(1)$. Определение. Идеал $I$ в $\mathbb{Z}[\omega]$ называется простым, если он делится ровно на два идеала: себя и (1). Упражнение 23. Докажите, что идеал $I$ прост тогда и только тогда, когда он максимален, т.е. когда единственный идеал больший его равен (1). Два идеала называются взаимнопростыми, если $I+J=(1)$. Упражнение 24. Докажите, что любые два различных простых идеала взаимнопросты. Задача 23. Проверьте, что данные идеалы просты в $\mathbb{Z}[\sqrt{-14}]$ : $$ p_{1}=(3,1+\sqrt{-14}), p_{2}=(3,1-\sqrt{-14}), p_{3}=(5,1+\sqrt{-14}), p_{4}=(5,1-\sqrt{-14}) $$ Задача 24. Докажите, что любой ненулевой идеал $I$ в $\mathbb{Z}[\omega]$ разлагается в произведение простых идеалов. Задача 25. Докажите, что если два идеала $I, J$ в $\mathbb{Z}[\omega]$ взаимнопросты и существует третий идеал $H$ такой, что $I$ делит $H J$, то $I$ делит $H$. Задача 26. а) Пусть $a \in \mathbb{Z}[\omega] \backslash 0$ и $(a) I=(a) J$ для двух идеалов $I, J \subset \mathbb{Z}[\omega]$. Докажите, что тогда $I=J$. б) Пусть для некотороых идеалов $I, H, J$ верно что $H \neq(0)$ и $H I=H J$. Докажите, что тогда $I=J$. Задача 27. Докажите, что разложение на простые множители из Задачи 24 единственно с точностью до перестановки множителей. ## Простые идеалы и простые числа Задача 28. а) Докажите, что любой идеал в $\mathbb{Z}[\sqrt{-5}]$ либо главный, либо имеет вид ( $1+$ $\sqrt{-5}) a, 2 a)$ для некоторого $a \in \mathbb{Q}[\sqrt{-5}]$. б) Докажите, что любой идеал в $\mathbb{Z}[\sqrt{-6}]$ либо главный, либо имеет вид $(\sqrt{-6} a, 2 a)$ для некоторого $a \in \mathbb{Q}[\sqrt{-6}]$. Задача 29. Докажите, что любой ненулевой идеал в $\mathbb{Z}[\omega]$ содержит ненулевое целое число. Задача 30. Докажите, что любой простой идеал $I$ в $\mathbb{Z}[\omega]$ содержит единственное простое число $p \in \mathbb{Z}, p>0$. Задача 31. Докажите, что для всякого простого идеала $I$ либо число $p=N(I)$ простое, либо оно квадрат простого числа $p>0$. Задача 32. Докажите, что простые числа $p$, определённые в двух предыдущих задачах, совпадают. Задача 33. Докажите, что для всякого простого числа $p \in \mathbb{Z}$ или идеал ( $p$ ) $\subset \mathbb{Z}[\omega]$ прост, или он равен произведению двух (не всегда различных) сопряжённых простых идеалов. Задача 34. Пусть $P_{\omega}(x)$ - это приведённый квадратный трёхчлен с целыми коэффициентами, для которого $P_{\omega}(\omega)=0$. Докажите, что в условиях предыдущей задачи первый случай имеет место тогда и только тогда, когда уравнение $P_{\omega}(x)=0$ не имеет решений по модулю $p$. ## Алгебраические числа В этой части мы бы хотим обсудить понятие алгебраического числа, а также те задачи, которые при этом возникают. В общем и целом этот материал часто присутствует в университетских курсах по теории чисел, но может быть усвоен и в старших классах школы. Определение. Комплексное число $\alpha \in \mathbb{C}$ называется алгебраическим, если оно является корнем ненулевого многочлена с рациональными коэффициентами. Алгебраическое число $\alpha \in \mathbb{C}$ называется целым, если оно является корнем приведённого многочлена с целыми коэффициентами. Множество алгебраических чисел обозначается $\overline{\mathbb{Q}}$. Множество целых алгебраических чисел обозначается $\overline{\mathbb{Z}}$. Задача 35. Если для рационального числа $a$ верно, что $a \in \overline{\mathbb{Z}}$, то $a \in \mathbb{Z}$. Задача 36* Докажите, что если $a, b \in \overline{\mathbb{Q}}$, то и $a \pm b \in \overline{\mathbb{Q}}, a b \in \overline{\mathbb{Q}}, a / b \in \overline{\mathbb{Q}}$ (в последнем случае считаем, что $b \neq 0$ ). Подсказка: попробуйте воспользоваться теоремой Виета. Задача 37* Пусть $b-$ это корень уравнения $a_{n} x^{n}+\ldots+a_{0}$, где $a_{0}, \ldots, a_{n} \in \overline{\mathbb{Q}}$. Докажите, что $b \in \overline{\mathbb{Q}}$. Задача 38. а) Определите деление с остатком в множестве многочленов от одной переменной с комплексными, вещественными и рациональными коэффициентами. б) Докажите единственность разложения на простые множители в множестве многочленов с рациональными коэффициентами. Задача 39. (Лемма Гаусса) Пусть $c_{g}$ - наибольший общий делитель коэффициентов многочлена $g \in \mathbb{Z}[x]$. Тогда для любых $g_{1}(x), g_{2}(x) \in \mathbb{Z}[x]$ выполнено $c_{g_{1} g_{2}}=c_{g_{1}} c_{g_{2}}$. Задача 40. Какие из конструкций и утверждений Задачи 38 применимы к многочленам с целыми коэффициентами от одной переменной? K многочленам с рациональными коэффициентами от двух переменных? Задача 41. Пусть $a$ - алгебраическое число. Пусть $P_{a}(x)$ - приведённый ненулевой многочлен наименьшей степени с рациональными коэффициентами, для которого $P_{a}(a)=0$. Докажите, что если $Q(a)=0$ для какого-то многочлена $Q(x)$, то $Q$ делится на $P_{a}$. Задача 42* Докажите, что если $a, b \in \overline{\mathbb{Z}}$, то и $a \pm b \in \overline{\mathbb{Z}}, a b \in \overline{\mathbb{Z}}$. Подсказка: попробуйте воспользоваться теоремой Виета. Задача 43* Пусть $a_{1}, \ldots a_{n}$ - алгебраические числа. Докажите, что если $a, b \in \mathbb{Q}\left[a_{1}, \ldots, a_{n}\right]$ и $b \neq 0$, то $\frac{a}{b} \in \mathbb{Q}\left[a_{1}, \ldots, a_{n}\right]$. Задача 44* Пусть $a$ - это целое алгебраическое число. Пусть $Q(x)$ - это приведённый многочлен наименьшей степени с рациональными коэффициентами, для которого $Q(a)=$ 0 . Докажите, что $Q(x)$ имеет целые коэффициенты. Задача 45* Пусть есть приведенный многочлен с целыми коэффициентами, такой что все его корни по модулю равны 1. Докажите, что тогда все его корни есть корни из 1. ## 3 Классы идеалов Определение. Для алгебраических $\alpha_{1}, \ldots, \alpha_{k}$ положим $\widetilde{\mathbb{Q}}:=\mathbb{Q}\left[a_{1}, \ldots, a_{k}\right]$ и $\widetilde{\mathbb{Z}}:=\overline{\mathbb{Z}} \cap \widetilde{\mathbb{Q}}$. Отметим, что любое подмножество в $\mathbb{C}$, замкнутое относительно сложения, вычитания и умножения называется колъцом. Определение. Непустое подмножество в $\widetilde{\mathbb{Z}}$ называется идеалом, если оно замкнуто относительно сложения, вычитания и умножения на элементы $\widetilde{\mathbb{Z}}$. Определение. Назовем идеалы $I, J \subseteq \widetilde{\mathbb{Z}}$ эквивалентными, если существуют ненулевые $\alpha, \beta \in \widetilde{\mathbb{Z}}$, такие что $(\alpha) I=(\beta) J$. Эквивалентность идеалов будем обозначать $I \sim J$. Задача 46. Проверьте, что $\sim$ является отношением эквивалентности. Определение. Классы эквивалентности идеалов будем называть классами идеалов. Задача 47. Докажите, что число классов идеалов равно 1 тогда и только тогда, когда все идеалы - главные. Задача 48. Проверьте, что если $I_{1} \sim I_{2}$ и $J_{1} \sim J_{2}$, то $I_{1} J_{1} \sim I_{2} J_{2}$. Задача 49. Опишите классы идеалов в $\mathbb{Z}[\sqrt{-5}]$ и $\mathbb{Z}[\sqrt{-6}]$. Как в них устроено умножение идеалов? Задача 50. Докажите, что если $I \subseteq(\alpha)$, то множество $(1 / \alpha) I$ - идеал в $\widetilde{\mathbb{Z}}$. В следующей серии задач обсуждается одно из фундаментальных утверждений алгебраической теории чисел. Это утверждение будет играть ключевую роль в доказательстве общего случая основной теоремы арифметики. Само доказательство обсуждается в следующей части проекта и может сдаваться в предположении что Теорема 6 уже доказана. Теорема 6. Число классов идеалов конечно. Определение. Назовем набор чисел $x_{1}, \ldots, x_{n} \in \widetilde{\mathbb{Q}}$ базисом над $\mathbb{Q}$, если любой элемент $a \in \widetilde{\mathbb{Q}}$ единственным образом представляется в виде $m_{1} x_{1}+\ldots+m_{n} x_{n}$, где $m_{1}, \ldots, m_{n} \in \mathbb{Q}$. Задача 51. Если $\alpha$ - алгебраическое, то в $\mathbb{Q}[\alpha]$ существует конечный $\mathbb{Q}$-базис. Задача 52. Докажите, что в $\widetilde{\mathbb{Q}}$ существует конечный базис над $\mathbb{Q}$. Задача 53. Для любого алгебраического $\alpha$ существует такое ненулевое $n \in \mathbb{Z}$, что $n \alpha-$ целое алгебраическое число. Задача 54. Пусть $I$ - идеал в $\widetilde{\mathbb{Z}}$. Докажите, что существует конечный набор $\alpha_{1}, \ldots, \alpha_{N} \in$ $I$, такой что любое $\alpha \in I$ представимо в виде $m_{1} \alpha_{1}+m_{2} \alpha_{2}+\ldots+m_{N} \alpha_{N}$ с целыми $m_{1}, \ldots, m_{N}$ (не обязательно единственным образом). Подсказка: докажите, что для фиксированного базиса в $\widetilde{\mathbb{Q}}$ коэффициенты $\alpha \in I$ не могут быть слишком маленькими. Задача 55. Докажите, что существует конечный набор $\alpha_{1}, \ldots, \alpha_{n} \in I$, для которого такое представление единственно. Подсказка: индукция по размеру базиса. Определение. Такие наборы будем называть целым базисом идеала $I$. Задача 56. Докажите, что любой целый базис $I$ является $\mathbb{Q}$-базисом $\widetilde{\mathbb{Q}}$. Определение. Обозначим $\widetilde{\mathbb{Z}} / I$ - множество классов эквивалентности элементов $\widetilde{\mathbb{Z}}$ по отношению эквивалентности $$ z_{1} \equiv z_{2} \quad \bmod I \Longleftrightarrow z_{1}-z_{2} \in I $$ Элементы $\widetilde{\mathbb{Z}} / I$ будем называть вычетами по модулю I. Задача 57. Проверьте, что это действительно отношение эквивалентности, и при $\widetilde{\mathbb{Z}}=\mathbb{Z}$ - определение вычета совпадает со стандартным. Задача 58. Чему равно количество элементов в $\widetilde{\mathbb{Z}} / I$ для квадратичных расширений? Задача 59. Докажите, что I содержит ненулевое целое число. Задача 60. Докажите, что число элементов в $\widetilde{\mathbb{Z}} / I$ конечно. Задача 61. Докажите, что существует только конечное число идеалов, содержащих данное $\alpha \in \widetilde{\mathbb{Z}}$. Зафиксируем целый базис $\alpha_{1}, \ldots, \alpha_{n}$ для идеала (1) $=\widetilde{\mathbb{Z}}$ и сопоставим каждому $\alpha \in \widetilde{\mathbb{Z}}$ набор целых чисел $\left(x_{1}, \ldots, x_{n}\right)$, таких что $\alpha=x_{1} \alpha_{1}+\ldots+x_{n} \alpha_{n}$. Положим $$ \|\alpha\|=\left|x_{1}\right|+\left|x_{2}\right|+\ldots+\left|x_{n}\right| $$ Задача 62. Пусть $\left\{\beta_{1}, \ldots, \beta_{m}\right\}$ - какой-то целый базиса идеала $I$. Докажите, что набор $\left\{\beta_{1}, \ldots, \beta_{m}\right\}$ является $\mathbb{Q}$-базисом в $\widetilde{\mathbb{Q}}$. Задача 63. Докажите, что существует такое натуральное $M_{1}$, что для любого ненулевого $\beta \in \widetilde{\mathbb{Z}}$ можно выбрать целый базис $\beta_{1}, \ldots \beta_{n}$ идеала ( $\beta$ ), такой что $\left\|\beta_{i}\right\|k$, такие что $I^{k} \sim I^{m}$. Задача 71. Докажите, что существует $\alpha \in \widetilde{\mathbb{Z}}$, такое что $I^{m-k}=(\alpha)$. В частности, для всякого идеала $I \subseteq \widetilde{\mathbb{Z}}$ существует идеал $J \subseteq \widetilde{\mathbb{Z}}$ и $\alpha \in \widetilde{\mathbb{Z}}$, такие что $I J=(\alpha)$. Задача 72. Докажите, что для любых двух идеалов $I, J$ в $\widetilde{\mathbb{Z}}$ - идеал $I$ делится на $J$ тогда и только тогда, когда $I$ содержится в $J$. Задача 73. Докажите, что любой идеал $I \subseteq \widetilde{\mathbb{Z}}$ представим в виде произведения конечного набора простых идеалов. Задача 74. Докажите, что если идеалы $I, J$ взаимнопросты и $J H \subseteq I$, то $H \subseteq I$. Задача 75. Докажите, что любые два различных простых идеала взаимно просты. Задача 76. Докажите, что если $I$ - простой идеал, и $I^{m} \subseteq J$, то $J=I^{k}$ для целого $k \leq m$. Задача 77. Докажите, что степени двух различных простых идеалов взаимно просты. Задача 78. Докажите однозначность разложения на множители в Задаче 73. ## Основная теорема арифметики и Великая Теорема Ферма Фиксируем простое число $p>2$ и обозначим через $\zeta_{p}$ комплексный корень $p$-ой степени из 1. Цель данного раздела доказать следующую теорему. Теорема 8. Пусть целые числа $x, y, z$ таковы что $x^{p}+y^{p}=z^{p}$ и число классов $\mathbb{Z}\left[\zeta_{p}\right]$ не делится на $p$. Тогда $x y z$ делится на $p$. Мы надеемся, что участники проекта смогут доказать Теорему 8, после того, как прорешают идущие ниже задачи. Упражнение 25. Докажите, что $1+\zeta_{p}+\zeta_{p}^{2}+. .+\zeta_{p}^{p-1}=0$. Задача 79. Найдите приведённый многочлен с целыми коэффициентами степени $p-1$, корнем которого является $1-\zeta_{p}$. Задача 80. Докажите, что все коэффициенты этого многочлена (кроме первого) делятся на $p$ и что этот многочлен неприводим. Упражнение 26. Докажите, что $\Sigma_{i=0}^{p-1} a_{i} \zeta_{p}^{i}=\sum_{i=0}^{p-1} b_{i} \zeta_{p}^{i}$ для рациональных чисел $$ a_{0}, . ., a_{p-1}, b_{0}, \ldots, b_{p-1} $$ если и только если $$ a_{0}-b_{0}=a_{1}-b_{1}=\ldots=a_{p-1}-b_{p-1} $$ Напомним что число $\gamma$ называется делителем единицы, если $\gamma$ и $1 / \gamma$ - целые алгебраические числа. Задача 81. Докажите, что $\frac{1-\zeta_{p}^{n}}{1-\zeta_{p}}$ - делитель единицы, если $n$ не делится на $p$. Задача 82. Докажите, что $\left(1-\zeta_{p}\right)^{p}=(p)$ (равенство идеалов). Упражнение 27. Докажите, что если $a \in \mathbb{Z}\left[\zeta_{p}\right]$ делится на $p$, то оно делится на $1-\zeta_{p}$. Пусть числа $a_{0}, \ldots, a_{p-1}$ рациональны, $a=a_{0}+. .+a_{p-1} \zeta_{p}^{p-1}$. Задача 83. Докажите, что все коэффициенты многочлена $$ P\left(a_{0}, \ldots, a_{p-1} ; x\right):=\Pi_{k=1}^{p-1}\left(x-\Sigma_{i=0}^{p-1} a_{i} \zeta_{p}^{k i}\right) $$ рассматриваемого как многочлен от $x$, также рациональны. Напомним, что для всякого алгебраического числа $a$ через $P_{a}(x)$ обозначается приведённый многочлен наименьшей степени, для которого $P_{a}(a)=0$. Задача 84*. Докажите, что $P\left(a_{0}, \ldots, a_{p-1} ; x\right)=P_{a}(x)^{d}$, где $d$ - целое положительное число. Задача 85. Если многочлен $P_{a}(x)$ имеет целые коэффициенты, то для любых целых чисел $1 \leq k \leq p-1, l \in \mathbb{Z}$, число $\Sigma_{i=0}^{p-1} a_{i} \zeta_{p}^{k i+l}$ является целым алгебраическим. Задача 86. Пусть многочлен $P_{a}(x)$ имеет целье коэффициенты. Докажите, что тогда $p\left(a_{i}-a_{j}\right) \in \mathbb{Z}$ для всех $0 \leq i, j \leq p-1$. Задача 87. Докажите, что число $1 /\left(1-\zeta_{p}\right)$ не является целым алгебраическим. Задача 88. Докажите что $\mathbb{Z}\left[\zeta_{p}\right]$ совпадает с множеством целых алгебраических чисел в $\mathbb{Q}\left[\zeta_{p}\right]$. Подсказка: постарайтесь найти применение Упражнению 27. Задача 89. Докажите, что для любого элемента $a \in \mathbb{Z}\left[\zeta_{p}\right]$ существует $b \in \mathbb{Z}$, такое что $a^{p}-b$ делится на $p$. Задача 90. а) Докажите, что $\sqrt{-1} \notin \mathbb{Z}\left[\zeta_{p}\right]$. б) Пусть $q \neq p, q \neq 2$ простое число, а $\zeta_{q}$ - комплексный корень $q$-ой степени из 1 . Докажите, что $\zeta_{q} \notin \mathbb{Z}\left[\zeta_{p}\right]$. в) Пусть $\zeta_{p^{2}}$ - корень $p^{2}$-ой степени из 1 , не являющийся корнем $p$-ой степени из 1 . Докажите, что $\zeta_{p^{2}} \notin \mathbb{Z}\left[\zeta_{p}\right]$. г) Найдите все корни из 1 в $\mathbb{Z}\left[\zeta_{p}\right]$. Задача 91* Докажите, что любой делитель единицы в $u \in \mathbb{Z}\left[\zeta_{p}\right]$ представляется в виде $\zeta_{p}^{i} v$, где $i \in \mathbb{Z}, v \in \mathbb{R}$. Пусть $x, y, z, p$ как в теореме 8 . Задача 92. Докажите что идеалы ( $x+\zeta_{p}^{i} y$ ) попарно взаимно просты при $0 \leq i \leq p-1$. Задача 93. Докажите Теорему 8. ## Список литературы [IR] K. Айерлэнд, М. Роузен, Классическое введение в современную теорию чисел, Мир, 1987. [Po] М. Постников, Теорема Ферма, Наука, 1978. [C1] K. Conrad, Factoring in quadratic fields, http://www.math.uconn.edu/ kconrad/blurbs/gradnumthy/quadraticgrad.pdf. [C2] K. Conrad, Ideal factorization, http://www.math.uconn.edu/ kconrad/blurbs/gradnumthy/idealfactor.pdf. [Mi] J. Milne, Algebraic number theory, http://jmilne.org/math/CourseNotes/ANT.pdf. [Wa] L. Washington, Introduction to Cyclotomic Fields, Graduate Texts in Mathematics 83, Springer-Verlag, 1996. [Go] А. Гончаров, Арифметика гауссовых чисел, Журнал "Квант"12 (1985), http://kvant.mccme.ru/1985/12/. [SS] В. Сендеров, А. Спивак, Сумми квадратов и целые гауссовы числа, Журнал "Квант"3 (1993), см. также http://kvant.mccme.ru/pdf/1999/03/. [ZSS] А. Заславский, А. Скопенков, М. Скопенков (редакторы), Элементы математики в задачах - через олимпиады и кружки к профессии, 2-ое изд., издательство МЦНМО, 2017. Задача 1. $(a+b i)+(c+d i)=(a+c)+(b+d) i$ $(a+b i)(c+d i)=(a c-b d)+(a d+b c) i$. Упражнение 1. Делители единицы имеют норму 1 и находятся перебором. Задача 2. Докажем деление с остатком по норме $N(a+b i)=(a+$ $b i)(a-b i)=a^{2}+b^{2}$ в целых гауссовых числах: если $x, y \in \mathbb{Z}[i]$ и $y \neq 0$, то существуют $r, t \in \mathbb{Z}[i]$, такие что $x=y r+t$ и $N(t)1$. Значит, $\operatorname{gcd}(x+i, x-i)=\operatorname{gcd}(2 i, x-i)=$ $\operatorname{gcd}(2 i,-i)=1$, то есть $x+i$ и $x-i$ взаимно просты. Значит, из единственности разложения на множители, $x+i=\varepsilon z^{n}$, где $z \in \mathbb{Z}[i]$, а $\varepsilon$ - делитель единицы в $\mathbb{Z}$, то есть $\varepsilon \in\{1,-1, i,-i\}$. Но $n$ не делится на 2 , так что после умножения $z$ на делитель единицы можно считать, что $x+i=z^{n}$. Пусть $z=a+b i$, тогда $i=\operatorname{Im} z^{n}=i\left(n a^{n-1} b-\binom{n}{3} a^{n-3} b^{3}+\ldots+(-1)^{\frac{n-1}{2}} b^{n}\right)$. Значит, 1 делится на $b$, то есть $b= \pm 1$. Заметим, что $a$ делится на 2 , т.к. иначе $z=a+b i \equiv 1+i \bmod 2$, поэтому $z^{2} \equiv(1+i)^{2} \equiv 0 \bmod 2$ и $x^{2}+1=z^{n} \equiv 0 \bmod 2$, но $x$ делится на 2 , противоречие. Если $a=0$, то $x+i= \pm i$, поэтому $x=0$ и $y=1$. Если $a \neq 0$, то $1 \equiv(-1)^{\frac{n-1}{2}} b^{n} \bmod 4$, поэтому $1=(-1)^{\frac{n-1}{2}} b^{n}$ и $0=$ $\binom{n}{n-2}+\ldots+(-1)^{\frac{n-5}{2}}\binom{n}{3} a^{n-5}+(-1)^{\frac{n-3}{2}} n a^{n-3}=\binom{n}{2}+\ldots+(-1)^{\frac{n-5}{2}}\binom{n}{n-3} a^{n-5}+$ $(-1)^{\frac{n-3}{2}}\binom{n}{n-1} a^{n-3}$. Докажем, что все слагаемые в этой сумме, кроме ( $\left.\begin{array}{l}n \\ 2\end{array}\right)$, делятся на степень 2 , большую, чем ( $\left.\begin{array}{l}n \\ 2\end{array}\right)$ - из этого очевидно следует, что равенство невозможно. Для этого заметим, что $$ \binom{n}{2 k} a^{2 k-2}=\binom{n}{2}\binom{n-2}{2 k-2} \frac{2 a^{2 k-2}}{(2 k-1) 2 k}=\binom{n}{2}\binom{n-2}{2 k-2} \frac{(a / 2)^{2 k-2}}{2 k-1} \frac{2^{2 k-2}}{k} $$ и $2^{2 k-2}>k$ при $k \geq 2$, из чего следует предыдущее утверждение. Значит, уравнение не имеет других решений, кроме $x=0, y=1$ при любом $n \geq 2$, и $x=0, y=-1$ при четном $n$. Задача 4. а) Если $p=2$, то $p=(1+i)(1-i)$, поэтому $p$ не простое в гауссовых целых числах. Если $p=4 k+3$, то предположим противное, т.е. $p$ не простое гауссово число. Тогда пусть $p=q_{1} q_{2} \ldots q_{n}$ - разложение на простые. Заметим, что $\bar{p}=p=\bar{q}_{1} \ldots \bar{q}_{n}$, поэтому $p$ делится на сопряженные к своим простым делителям. Заметим, что сопряженное к простому также простое (иначе его разложение можно снова сопрячь), и в нашем случае оно отличается от исходного (с точностью до домножения на делитель единицы). Действительно, иначе при $q=a+b i$ число $q / \bar{q}=(a+b i) /(a-b i)=$ $\left(\left(a^{2}-b^{2}\right)+2 a b i\right) /\left(a^{2}+b^{2}\right)$ должно быть целым гауссовым, что возможно только в случаях $a= \pm b$ и $a b=0$ (т.к. $|2 a b| \leq a^{2}+b^{2}$ ). В первом случае $q$ делится на $1+i$, поэтому $p \bar{p}=p^{2}$ делится на $(1+i)(1-i)=2$, противоречие. Во втором случае можно считать, что $q$ - натуральное (с точностью до домножения на делитель единицы), но $p$ простое в целых числах, поэтому $q=p$ и $p$ - простое в целых гауссовых, противоречие. Значит, $q$ и $\bar{q}$ - различные простые ( $q$ - простой делитель $p$ в целых гауссовых), и $p$ делится на оба. Значит, $p$ делится на $q \bar{q}=a^{2}+b^{2}$, и т.к. $p$ - простое в целых числах, то $p=a^{2}+b^{2}$. Но как известно, простое $p$ вида $4 k+3$ нельзя представить в виде суммы двух квадратов (посмотрим на остатки по модулю 4), противоречие. Если $p=4 k+1$, то (как известно) -1 является вычетом по модулю $p$, поэтому существует натуральное $n$ такое, что $n^{2}+1=(n+i)(n-i)$ делится на $p$. Если бы $p$ было простым гауссовым, то $n+i$ или $n-i$ делилось бы на $p$, что очевидно не верно. Значит, $p$ не простое гауссово. б) Пусть $n=a^{2}+b^{2}=(a+b i)(a-b i)$ и $m=c^{2}+d^{2}=(c+d i)(c-d i)$, тогда $n m=((a+b i)(c+d i))((a-b i)(c-d i))=((a c-b d)+(a d+b c) i)((a c-$ $b d)-(a d+b c) i))=(a c-b d)^{2}+(a d+b c)^{2}$. в) Рассмотрим минимальное такое $n$, что $n=x^{2}+y^{2}$ для целых $x$ и $y$, и при этом не удовлетворяет условию теоремы 2 . Тогда $n$ делится на какое-то простое $p=4 k+3$ в нечетной степени. Но т.к. -1 невычет по модулю $p$, то из того, что $x^{2}+y^{2}$ делится на $p$, следует, что $x$ и $y$ делятся на $p$. Значит, $n$ делится на $p^{2}$ и $n / p^{2}=(x / p)^{2}+(y / p)^{2}$, причем $n / p^{2}$ делится на $p$ в нечетной степени, что противоречит минимальности $n$. Значит, если натуральное $n=x^{2}+y^{2}$, то оно удовлетворяет условию теоремы 2. Обратно, если $n$ удовлетворяет условию теоремы 2 , то из пункта б) достаточно показать, что 2 , простое $p=4 k+1$ и $p^{2}$ для простого $p=4 k+3$ - представимы в виде суммы двух квадратов. Действительно, $2=1^{2}+1^{2}$, $p^{2}=p^{2}+0^{2}$ для любого $p$, а $p=4 k+1$ - составное в целых гауссовых, поэтому делится на гауссово простое $q=a+b i$, причем (аналогично пункту а)) $q \neq \pm 1 \pm i$ и $a b \neq 0$, поэтому $p$ делится на $q \bar{q}=a^{2}+b^{2}$. Значит, $p=a^{2}+b^{2}$ из простоты $p$ в целых числах. Задача 5. Будем считать, что $n$ удовлетворяет условию теоремы 2 (т.е. представляется хотя бы одним способом). Тогда, как видно из решения задачи 4 , для любого представления $n$ в виде суммы $x^{2}+y^{2}-$ мы можем сократить на все простые вида $4 k+3$, поэтому можно считать, что $n$ не делится на простые вида $4 k+3$ (число представлений от этого не изменится). Пусть $n=2^{\alpha_{0}} p_{1}^{\alpha_{1}} \ldots p_{m}^{\alpha_{m}}$ - разложение на простые в целых числах, тогда, как видно из решения задачи 4 , его разложения в целых гауссовых будет иметь вид $n=(1+i)^{\alpha_{0}}(1-i)^{\alpha_{0}} q_{1}^{\alpha_{1}}\left(\bar{q}_{1}\right)^{\alpha_{1}} \ldots q_{m}^{\alpha_{m}}\left(\bar{q}_{m}\right)^{\alpha_{m}}$, где $q_{i}$ - простой делитель $p_{i}$. Каждому представлению $n$ в виде суммы двух квадратов соответствует разложение $n=(x+y i)(x-y i)=z \bar{z}$, где $z=x+$ yi определено с точность до домножения на делители единицы и сопряжения. Количество способов составить $z$ с точностью до делителей единицы, как видно из разложения $n$ на простые гауссовы, равно $\left(\alpha_{1}+1\right)\left(\alpha_{2}+1\right) \ldots\left(\alpha_{m}+1\right)$ (т.к. $1+i$ и $1-i$ - одинаковые простые). При сопряжении $z$ совпадает с собой (с точностью до делителя единицы) тогда и только тогда, когда $z$ делится на каждое простое в степени, в 2 раза меньшей степени вхождения в $n$. Такое возможно только в том случае, когда все $\alpha_{1}, \ldots, \alpha_{m}$ - четные, т.е. $\left(\alpha_{1}+1\right)\left(\alpha_{2}+1\right) \ldots\left(\alpha_{m}+1\right)-$ нечетное. Значит, окончательная формула количества представлений $n$ в виде суммы двух квадратов $-\left\lfloor\left(\left(\alpha_{1}+1\right)\left(\alpha_{2}+1\right) \ldots\left(\alpha_{m}+1\right)+1\right) / 2\right\rfloor$. Задача 6. Докажем, что уравнение $x^{3}+y^{3}+z^{3}=0$ с $x y z \neq 0$ не имеет решений в целых числах. Предположим противное, тогда после сокращения на $\operatorname{gcd}(x, y, z)$ можем считать, что $x, y, z$ взаимно просты в совокупности (а значит, и попарно). Если $x y z$ не делится на 3 , то $x^{3}, y^{3}, z^{3} \equiv \pm 1$ $\bmod 9$, поэтому $x^{3}+y^{3}+z^{3} \not \equiv 0 \bmod 9$. Значит, одно из $x, y, z$ делится на 3 (из взаимной простоты - ровно одно) - без ограничения общности $z$ делится на 3. Заменяя $z$ на $-z$, получаем уравнение $x^{3}+y^{3}=z^{3}$, где $z$ делится на 3 . Далее будем работать в $\mathbb{Z}[\xi]$, где $\xi=\frac{-1+\sqrt{-3}}{2}$, и обозначим $\pi=1-\xi$ - простое число, такое что $\pi \bar{\pi}=-\xi^{2} \pi^{2}=3$. Докажем, что если для взаимно простых $x, y, z \in \mathbb{Z}[\xi]$ выполнено $x^{3}+$ $y^{3}=u z^{3}$, где $u$ - делитель единицы, и $z$ делится $\pi$, то $z$ делится на $\pi^{2}$. Действительно, $x^{3}+y^{3}=(x+y)(x+\xi y)\left(x+\xi^{2} y\right)=(x+y)(x+y-\pi y)(x+$ $y+\pi y-\xi^{2} \pi^{2} y$ ), поэтому из того, что $x^{3}+y^{3}=u z^{3}$ делится на $\pi$, следует, что $x+y$ делится на $\pi$, значит, все 3 множителя в разложении делятся на $\pi$, причем $x+y \not \equiv x+y-\pi y \not \equiv x+y+\pi y \bmod \pi^{2}$ (т.к. $y$ не делится на $\pi$ ), и каждое число в $\mathbb{Z}[\xi]$ очевидно сравнимо с 0,1 или 2 по модулю $\pi=1-\xi$. Значит, один из этих множителей делится на $\pi^{2}$, поэтому $u z^{3}$ делится на $\pi^{4}$. Значит, $z$ делится на $\pi^{2}$. Теперь докажем, что если для взаимно простых $x, y, z \in \mathbb{Z}[\xi]$ выполнено $x^{3}+y^{3}=u z^{3}$, где $u$ - делитель единицы, $z$ делится $\pi^{k}$ и не делится на $\pi^{k+1}$ с $k \geq 2$, то существуют взаимно простые $x_{1}, y_{1}, z_{1} \in \mathbb{Z}[\xi]$ с $x_{1}^{3}+y_{1}^{3}=u_{1} z_{1}^{3}$, где $u_{1}$ - делитель единицы, $z$ делится $\pi^{k-1}$ и не делится на $\pi^{k}$. Разложением $x^{3}+y^{3}=(x+y)(x+\xi y)\left(x+\xi^{2} y\right)=(x+y)(x+y-$ $\pi y)\left(x+y+\pi y-\xi^{2} \pi^{2} y\right)=u z^{3}$ мы снова получаем, что все 3 множителя в разложении делятся на $\pi$, причем ровно один из них делится на $\pi^{2}$ (а значит, на $\pi^{3 k-2}$ ). После домножения $y$ на $\xi$ или $\xi^{2}$ можно считать, что $x+y$ делится на $\pi^{3 k-2}$. Из единственности разложения получаем, что $x+y=u_{1} \pi^{3 k-2} a^{3}$, $x+\xi y=u_{2} \pi b^{3}, x+\xi^{2} y=u_{3} \pi c^{3}$, где $u_{i}-$ делители единицы, и $a, b, c$ не делятся на $\pi$. Из равенства $0=(x+y)+\xi(x+\xi y)+\xi^{2}\left(x+\xi^{2} y\right)$ получаем $u_{1} \pi^{3 k-2} a^{3}+\xi u_{2} \pi b^{3}+\xi^{2} u_{3} \pi c^{3}=0$. После сокращения уравнение можно привести к виду $b^{3}+u_{4} c^{3}=u_{5}\left(a \pi^{k-1}\right)^{3}$, где $u_{4}$ и $u_{5}$ - делители единицы. Докажем, что $u_{4}= \pm 1$. Действительно, заметим, что если $b=n+m \xi$, где $n, m$ целые числа, то $b^{3}=(n+m \xi)^{3} \equiv n^{3}+m^{3} \equiv \pm 1 \bmod 3$, т.к. $b$ не делится на $\pi$. Аналогично $c^{3} \equiv \pm 1 \bmod 3$, и $\left(\pi^{k-1}\right)^{3}$ делится на 3 из $k \geq 2$. Значит, $\pm 1 \pm u_{4}$ делится на 3 , поэтому $u_{4}= \pm 1$. После замены $c$ на $u_{4} c$ получаем $b^{3}+c^{3}=u_{5}\left(a \pi^{k-1}\right)^{3}$, откуда мы получаем нужные $x_{1}, y_{1}, z_{1}$. Теперь можно рассмотреть целочисленное решение $x^{3}+y^{3}=z^{3}$ с $z$, делящимся на 3 , и спуском по степени вхождения $\pi$ в $z$ получить противоречие, т.к. $z$ делится на $\pi^{2}$. Упражнение 2. Имеем $x^{3}-1=(x-1)\left(x^{2}+x+1\right)=0$ и решаем квадратное уравнение. Задача 7. Заметим, что для $a=x+y i \in \mathbb{Z}[\xi]$ стандартная норма $N(a)=x^{2}+y^{2}$ будет целой, т.к. $a=(z+\sqrt{-3} t) / 2$, где $z$ и $t$ - одной четности. Для ненулевого $a \in \mathbb{Z}\left[\frac{1+\sqrt{-3}}{2}\right]$ числа, кратные $a$, образуют решетку в плоскости комплексных чисел, разбивающую её на правильные треугольники со стороной $|a|$. Для любой точки в таком треугольнике квадрат расстояния от неё до любой вершины (кроме остальных в том случае, когда точка совпадает с вершиной) меньше $|a|^{2}=N(a)$, откуда следует деление с остатком по норме комплексных чисел (чтобы поделить $b$ на $a$ с остатком, надо вычесть из $b$ ближайшее к нему на комплексной плоскости число вида $a c$ ). Аналогично задаче 2 , отсюда мы получаем основную теорему арифметики для $\mathbb{Z}[\xi]$. Задача 8. Для делителей единицы в $\mathbb{Z}[\xi]$ норма должна быть 1 (т.к. у обратного норма должна быть целой). Для $a=(x+\sqrt{-3} y) / 2 \in \mathbb{Z}[\xi]-$ $N(a)=\left(x^{2}+3 y^{2}\right) / 4$, поэтому если $a-$ делитель единицы, то $-2 \leq x \leq 2$ и $-1 \leq y \leq 1$. Перебором получаем, что все делители единицы в $\mathbb{Z}[\xi]-$ это $\pm 1, \pm \xi$ и $\pm \xi^{2}$. Упражнение 3. а) Замкнутость относительно умножения следует из равенства $(x+y \sqrt{d})(z+t \sqrt{d})=(x z+d y t)+(x t+y z) \sqrt{d}$. б) $\frac{x+y \sqrt{d}}{z+t \sqrt{d}}=\frac{(x z-d y t)+(y z-x t) \sqrt{d}}{z^{2}-d t^{2}}$. Упражнение 4. $\xi$ является целым в $\mathbb{Q}[\sqrt{-3}] \cdot \frac{1+i}{2}$ не является целым в $Z[i]$. Упражнение 5. Очевидно. Упражнение 6. a) $x^{2}-\operatorname{Tr}(\alpha) x+N(\alpha)=(x-\alpha)(x-\bar{\alpha})$. Подставляя в $x=\alpha$, получаем ноль. б) Вычисление. Задача 9. По предыдущему упражнению $x+y \sqrt{d}$ цело в $\mathbb{Q}[\sqrt{d}]$ тогда и только тогда, когда числа $2 x$ и $x^{2}-d y^{2}$ целы. Легко проверить, что для элементов $Z[\omega]$ это условие выполнено. Пусть числа $2 x$ и $x^{2}-d y^{2}$ целы. Если $x$ целое, то $d y^{2}$ целое, откуда $y$ целое и $x+y \sqrt{d} \in Z[\omega]$. Если $x=n / 2$, где $n$ нечетно, то $4 d y^{2}$ целое, откуда $y=m / 2$, где $m$ целое. При этом $d m^{2} \equiv n^{2} \equiv 1 \bmod 4$, откуда $m$ нечетно и $d \equiv 1 \bmod 4$. Тогда $x+y \sqrt{d}=m \omega+(n-m) / 2 \in \mathbb{Z}[\omega]$. Упражнение 7. Если $N(\gamma)= \pm 1$, то $\frac{1}{\gamma}= \pm \bar{\gamma} \in \mathbb{Z}[\omega]$. Если $\frac{1}{\gamma} \in \mathbb{Z}[\omega]$, то $1=N(1)=N\left(\gamma \frac{1}{\gamma}\right)=N(\gamma) N\left(\frac{1}{\gamma}\right)$, откуда $N(\gamma)= \pm 1$, поскольку $N\left(\frac{1}{\gamma}\right) \in \mathbb{Z}$. Упражнение 8. Предположим, что $\gamma$ не просто: $\gamma=a b, a, b \in \mathbb{Z}[\omega]$, $N(a) \neq \pm 1, N(b) \neq \pm 1$. Тогда $|N(\gamma)|=|N(a)||N(b)|$, откуда одно из (натуральных) чисел $|N(a)|$ и $|N(b)|$ равно 1. Противоречие. Покажем, что $5 \in Z[\sqrt{3}]$ простое, хотя $N(5)=25$ - составное число. Пусть $5=a b$, где $a, b \in \mathbb{Z}[\sqrt{3}], N(a) \neq \pm 1, N(b) \neq \pm 1$. Тогда $N(a) N(b)=$ 25 , откуда $N(a)= \pm 5$. Если $a=x+y \sqrt{3}$, то $x^{2}-3 y^{2}= \pm 5$. Это уравнение не имеет решений по модулю 3 . Упражнение 9. Индукция по $|N(\gamma)|$. Задача 10. Поскольку $N(3)=9, N(5)=25, N(1+\sqrt{-14})=N(1-$ $\sqrt{-14})=15$, норма неотрицательна, и норма произведения равна произведению норм, то достаточно показать, что не существует числа $\alpha \in$ $\mathbb{Z}[\sqrt{-14}]$ с $N(\alpha)=3$ или $N(\alpha)=5$. Пусть $\alpha=x+y \sqrt{-14}$, где $x, y-$ целые. А уравнения $x^{2}+14 y^{2}=3$ и $x^{2}+14 y^{2}=5$ не имеют решений в целых числах, поскольку $x^{2}+14 y^{2}>5>3$ при $y \neq 0$. Задача 11. а) Для ненулевого $b \in \mathbb{Z}[\sqrt{-2}]$ числа, кратные $b$, образуют решетку в плоскости комплексных чисел, разбивающую её на прямоугольники со сторонами $|b|$ и $\sqrt{2}|b|$. Для любой точки внутри такого прямоугольника существует вершина, для которой квадрат расстояния от неё точки не больше $\left(\frac{1}{2}|b|\right)^{2}+\left(\frac{1}{2} \sqrt{2}|b|\right)^{2}=\frac{3}{4}|b|^{2}1$ получаем $b=0$, значит $a= \pm 1$, т.е. все делители единицы - это $\pm 1$. Если же $d=1$, то $-1 \leq a, b \leq 1$, откуда перебором получаем, что все делители единицы - это $\pm 1$ и $\pm i$. в) Задача 8 . г) Докажем, что все делители единицы имеют вид $\pm(1+\sqrt{2})^{n}$ для всех целых $n$. Т.к. $(1+\sqrt{2})(1-\sqrt{2})=-1$, то все такие числа - делители единицы. Предположим, что существует какой-то другой делитель единицы $a+b \sqrt{2}$. После домножения на -1 и сопряжения можно считать, что $a, b>0$ (если $a b=0$, то $a= \pm 1$ и $b=0$ из $a^{2}-2 b^{2}= \pm 1$ ). Рассмотрим минимальный такой делитель единицы $a+b \sqrt{2}$ (с $a, b \geq 0$ и $\left.a+b \sqrt{2} \neq \pm(1+\sqrt{2})^{n}\right)$. Тогда $(a+b \sqrt{2})(\sqrt{2}-1)=(2 b-a)+(a-b) \sqrt{2}$. Докажем, что $a>b$ и $2 b>a$. Действительно, заметим, что если $b=1$, то $a^{2}=2 \pm 1$, откуда $a=1$ и $a+b \sqrt{2}=1+\sqrt{2}$, противоречие. Значит, $b>1$, поэтому $a^{2}=2 b^{2} \pm 1>b^{2}$, откуда $a>b$. Если $a=1$, то $2 b^{2}=1 \pm 1$, откуда $b=1$ и $a+b \sqrt{2}=1+\sqrt{2}$, противоречие. Значит, $a>1$, поэтому $(2 b)^{2}=2\left(a^{2} \pm 1\right)>a^{2}$, откуда $2 b>a$. Отсюда получаем $0<2 b-a2$ делятся на степень двойки большую, чем $4\binom{n}{5}$, из чего будет следовать противоречие. Для $\frac{(n+1)(n-1)(n-3)}{3}$ это очевидно, т.к. $n$ на делится на 2 и $4\binom{n}{5}=\frac{n(n-1)(n-2)(n-3)(n-4)}{30}$. Перепишем второй тип слагаемых: $$ 2^{k}\binom{n}{2 k+1}=4\binom{n}{5}\binom{n-5}{2 k-4} \frac{15 * 2^{k-1}}{(2 k-3)(2 k-1)(2 k+1)(k-1) k} $$ Осталось заметить, что $2^{k-1}>k>k-1$ при $k>2$, поэтому $2^{k-1} /(k(k-1))$ делится на 2 , из чего следует наше утверждение. б) Если $n$ четно, то $7=\left(2^{n / 2}+k\right)\left(2^{n / 2}-k\right)$, поэтому $2^{n / 2}=4$ и $k= \pm 3$, т.е. $n=4, k= \pm 3$ - единственное решение при четном $n$. Далее считаем $n$ нечетным. Разложив обе части уравнения $2^{n-2}=\left(k^{2}+7\right) / 4$ на множители в $\mathbb{Z}\left[\frac{1+\sqrt{-7}}{2}\right]$, получаем $$ ((1+\sqrt{-7}) / 2)^{n-2}((1-\sqrt{-7}) / 2)^{n-2}=((k+\sqrt{-7}) / 2)((k-\sqrt{-7}) / 2) $$ где $(1+\sqrt{-7}) / 2$ и $(1-\sqrt{-7}) / 2-$ различные простые, $(k+\sqrt{-7}) / 2$ и $(k-\sqrt{-7}) / 2$ - взаимно просты. Обозначим $m=n-2$, тогда из выполнения основной теоремы арифметики для $\mathbb{Z}\left[\frac{1+\sqrt{-7}}{2}\right]$ получаем, что $((1+\sqrt{-7}) / 2)^{m}=( \pm k \pm \sqrt{-7}) / 2$. Если $m=1$, то $n=3$ и $k= \pm 1$. Докажем, что если $m>1$, то $((1+\sqrt{-7}) / 2)^{m}=( \pm k-\sqrt{-7}) / 2$. Действительно, иначе $((1+\sqrt{-7}) / 2)^{m}-((1-\sqrt{-7}) / 2)^{m}=\sqrt{-7}$, и по модулю $(-3-\sqrt{-7}) / 2=$ $((1-\sqrt{-7}) / 2)^{2}$ получаем, что $((1+\sqrt{-7}) / 2)^{m}-((1-\sqrt{-7}) / 2)^{m} \equiv(-1)^{m} \equiv$ $-1 \equiv \sqrt{-7}$, поэтому $\left.1+\sqrt{-7}=((1+\sqrt{-7}) / 2)^{2}(1-\sqrt{-7}) / 2\right)$ делится на $((1-\sqrt{-7}) / 2)^{2}$, противоречие. Значит, $((1+\sqrt{-7}) / 2)^{m}=( \pm k-\sqrt{-7}) / 2$, и приравнивая коэффициенты при $\sqrt{-7}$ по модулю 7 , получаем $-2^{m-1} \equiv m \bmod 7$. Т.к. 2 является вычетом по модулю 7 , то $m \equiv 3,5,6 \bmod 7$, при этом $m \equiv 0,2,1 \bmod 3$ соответственно. Значит, $m \equiv 3,5,13 \bmod 21$, при этом $m=3,5,13$ подходят - им соответствуют решения $n=5, k= \pm 5 ; n=7, k= \pm 11$; $n=15, k= \pm 181$. Докажем, что это все решения. Предположим противное, тогда $m \equiv m_{0} \bmod 21$, где $m_{0}=3,5$ или 13 , и $m>m_{0}$. Пусть $r=m-m_{0}$ делится на $7^{\alpha}$ и не делится на $7^{\alpha+1}$. Тогда по модулю $7^{\alpha+1}$ получаем $$ (1+\sqrt{-7})^{r}=1+r \sqrt{-7}+\ldots+(\sqrt{-7})^{l}\binom{r}{l}+\ldots+(\sqrt{-7})^{r} \equiv 1+r \sqrt{-7} $$ т.к. $l$ ! делится на 7 в степени, меньшей $l / 6=l / 7+l / 49+\ldots$.. Также $2^{r} \equiv(8)^{r / 3} \equiv 1 \bmod 7^{\alpha+1}$. Значит, $$ \begin{gathered} ((1+\sqrt{-7}) / 2)^{m}=((1+\sqrt{-7}) / 2)^{m_{0}}((1+\sqrt{-7}) / 2)^{r} \equiv\left(\left(k_{0}-\sqrt{-7}\right) / 2\right)(1+r \sqrt{-7}) \equiv \\ \equiv\left(\left(k_{0}+7 r\right)+\left(k_{0} r-1\right) \sqrt{-7}\right) / 2 \equiv(k-\sqrt{-7}) / 2 \bmod 7^{\alpha+1} \end{gathered} $$ поэтому $k_{0} r$ делится на $7^{\alpha+1}$. Но тогда $k_{0}$ делится на 7 , противоречие (т.к. иначе $2^{n}$ делится на 7 ). Значит, других решений нет. Задача 16. Мы будем использовать задачи 27 и 28. a) $\mathrm{B} \mathbb{Z}[\sqrt{-5}]$ имеется разложение $(x+\sqrt{-5})(x-\sqrt{-5})=y^{3}$. Т.е. произведение идеалов $(x+\sqrt{-5})$ и $(x-\sqrt{-5})$ является кубом идеала ( $y$ ). Если идеалы $(x+\sqrt{-5})$ и $(x-\sqrt{-5}$ ) не взаимно просты, то они имеют общий простой делитель $I$. Тогда идеал, порожденный $2 \sqrt{-5}=(x+\sqrt{-5})-(x-\sqrt{-5})$ делит $I$. Разложим $(2 \sqrt{-5})=(2,1+\sqrt{-5})^{2}(\sqrt{-5})$. Т.е. $I=(2,1+\sqrt{-5})$ или $I=(\sqrt{-5})$. В первом случае $x$ нечетно, откуда $x^{2}+5=y^{3}$ делится на 2 , но не делится на 4 , противоречие. Во втором случае $x$ делится на 5 , откуда $x^{2}+5=y^{3}$ делится на 5 , но не делится на 25 , противоречие. Значит, идеалы $(x+\sqrt{-5})$ и $(x-\sqrt{-5})$ взаимно просты и каждый из них является кубом некоторого идеала. Используя задачу 28 нетрудно проверить, что куб неглавного идеала будет неглавным, откуда $(x+\sqrt{-5}$ ) - куб главного идеала. Т.е. $(x+\sqrt{-5})=(a+b \sqrt{-5})^{3}$, поскольку делители единицы в $\mathbb{Z}[\sqrt{-5}]$ имеют вид $\pm 1$ и являются кубами. Раскрывая скобки и приравнивая коэффициенты при $\sqrt{-5}$ получаем $\left(3 a^{2}-5 b^{2}\right) b=3 a^{2} b-5 b^{3}=1$. Отсюда $b= \pm 1$. Легко видеть, что оба случая невозможны. Следовательно, уравнение не имеет решений. б) Заменяя $x$ на $z-1$ получаем уравнение $z^{2}+6=y^{3}$. Решений нет, доказательство аналогично пункту а). в) Разложим $5 x^{2}+1=(1+x \sqrt{-5})(1-x \sqrt{-5})=y^{3}$. Аналогично пункту a) $x$ четно и идеалы $(1+x \sqrt{-5})$ и $(1-x \sqrt{-5})$ взаимно просты. Тогда $(1+$ $x \sqrt{-5})=(a+b \sqrt{-5})^{3}$. Раскрывая скобки и приравнивая вещественные части получаем $a\left(a^{2}-15 b^{2}\right)=a^{3}-15 a b^{2}=1$. Если $a=1$, то $b=0$, откуда $x=0, y=1$. Случай $a=-1$ невозможен. Т.е. $x=0, y=1$. г) Заменяя $x$ на $z-1$ получаем уравнение $6 z^{2}+1=y^{3}$. Решение единственно: $x=0, y=1$, доказательство аналогично пункту в). д) Приведем решение, не использующее идеалы. Похожие решения есть и для пунктов а,б). Прибавим 8 и разложим на множители: $x^{2}+2=(y+2)\left(y^{2}-2 y+4\right)$. Рассмотрев остатки по модулю 4 получим, что $x$ нечетно, откуда $x^{2}+$ $2 \equiv 3 \bmod 8$. Используем, что -2 является квадратичным невычетом по модулю простого $p$ тогда и только тогда, когда $p \equiv 1$ или $3 \bmod 8$ (это следует, например, из квадратичного закона взаимности). Тогда все простые делители $x^{2}+2$ имеют вид $8 k+1$ или $8 k+3$, т.е. все делители $x^{2}+2=(y+2)\left(y^{2}-2 y+4\right)$ дают остатки 1 или 3 при делении на 8. Т.к. $y+2$ дает остаток 1 или 3 по модулю 8 , то $y$ дает остаток $\pm 1$. Из этих двух случаев $\left(y^{2}-2 y+4\right)$ дает остаток 1 или 3 только если $y \equiv 1 \bmod 8$. Отсюда $x^{2}+2=y^{3}+8 \equiv 1 \bmod 8$. Противоречие. Упражнение 10. Очевидно. Упражнение 11. Если $a, b \in I \cap J$, то $a+b \in I$, т.к. $a, b \in I$; аналогично $a+b \in J$. Отсюда $a+b \in I \cap J$. Если $a \in I \cap J, b \in \mathbb{Z}$, то $a b \in I \cap J$ по аналогичной причине. Задача 17. а,б) Следует из линейного представления НОД. в) Пусть $d$ - наименьшее натуральное число в идеале $I$ (если такого нет, то $I=(0))$. Если $I \neq(d)$, то $I$ содержит число $a$, не делящееся на $d$. Остаток от деления $a$ на $d$ лежит в $I$ и меньше $d$. Противоречие. Упражнение 12. Заметим, что ( $\alpha$ ) состоит из чисел вида $\alpha x$, где $x \in \mathbb{Z}[\omega]$, а также $(\alpha) \subset(\beta) \Leftrightarrow \alpha \in(\beta) \Leftrightarrow \alpha / \beta \in \mathbb{Z}[\omega]$. Имеем $(\alpha)=(\beta) \Leftrightarrow \alpha \in(\beta)$ и $\beta \in(\alpha) \Leftrightarrow \alpha / \beta \in \mathbb{Z}[\omega]$ и $\beta / \alpha \in \mathbb{Z}[\omega] \Leftrightarrow \alpha / \beta$ - делитель единицы Упражнение 13. Очевидно. Задача 18. Заметим, что $(2, \sqrt{-14})=\{x+y \sqrt{-14} \mid x$ четно, $y \in$ $\mathbb{Z}\}$ (проверьте!) Отсюда имеем, что, $1 \notin(2, \sqrt{-14}))$ и $(2, \sqrt{-14}) \neq(2)$. Предположим, что $(2, \sqrt{-14})=(\alpha)$. Тогда $2=\alpha x, x \in \mathbb{Z}[\sqrt{-14}]$, откуда $4=N(2)=N(\alpha) N(x)$. Если $N(\alpha)=1$, то $1=\alpha \bar{\alpha} \in(\alpha)$, т.е. $(1) \in(\alpha)=(2, \sqrt{-14})$, противоречие. $N(\alpha) \neq 2$, т.к. уравнение $x^{2}+14 y^{2}=2$ не имеет решений в целых числах. Если $N(\alpha)=4$, то $N(\sqrt{-14})=14$ не делится на $N(\alpha)$, т.е. $\sqrt{-14} \notin(\alpha)$. Задача 19. Пусть $\alpha$ - наименьшее натуральное число в $I, \beta=$ $a+b \omega$ - число с наименьшим положительным коэффициентом $b$ при $\omega$. Несложно проверить, что $I=(\alpha, \beta)$. Упражнения 14-15. Вычисления, не приводим. Упражнение 16. Из упражнения 13 достаточно найти ответ в $\mathbb{Z}$ : (360), (2), (180). Упражнение 17. Вычисления, мы приведем доказательство для $I+$ $J$. Если $a, b \in I+J$, то $a=i_{1}+j_{1}, b=i_{2}+j_{2}$, где $i_{1}, i_{2} \in I, j_{1}, j_{2} \in J$. Тогда $a+b=\left(i_{1}+i_{2}\right)+\left(j_{1}+j_{2}\right) \in I+J$, т.к. $i_{1}+i_{2} \in I, j_{1}+j_{2} \in J$. Для $c \in \mathbb{Z}[\omega]$ имеем $a c=\left(i_{1}+j_{1}\right) c=i_{1} c+j_{1} c \in I+J$, т.к. $i_{1} c \in I, j_{1} c \in J$. Задача 20. Поскольку $N\left(a_{i}\right)$ и $\operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)$ - целье числа, то имеем $\left(N\left(a_{i}\right), \operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)\right)_{1 \leq i, j \leq n}=(a)$, где $a=\operatorname{HOД}\left(N\left(a_{i}\right), \operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)\right)$. Заметим, что $N\left(a_{i}\right)$ и $\operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)$ делятся на $a$. Т.к. $N\left(a_{i}\right)=a_{i} \bar{a}_{i}$ и $\operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)=a_{i} \bar{a}_{j}+a_{j} \bar{a}_{i}$, то $$ (a)=\left(N\left(a_{i}\right), \operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)\right)_{1 \leq i, j \leq n} \subset\left(a_{1}, \ldots, a_{n}\right)\left(\bar{a}_{1}, \ldots, \bar{a}_{n}\right) $$ Достаточно доказать, что $a_{i} \bar{a}_{j} \in(a)$ для всех $1 \leq i, j \leq n$, т.е. что $a_{i} \bar{a}_{j} / a$ - целое алгебраическое. По упражнению 6 достаточно показать, что его норма и след целье. $N\left(a_{i} \bar{a}_{j} / a\right)=\frac{a_{i} \bar{a}_{j} a_{j} \bar{a}_{i}}{a^{2}}=\frac{N\left(a_{i}\right) N\left(a_{j}\right)}{a^{2}}$ целое, т.к. $N\left(a_{i}\right)$ и $N\left(a_{j}\right)$ делятся на $a \cdot \operatorname{Tr}\left(a_{i} \bar{a}_{j} / a\right)=\operatorname{Tr}\left(a_{i} \bar{a}_{j}\right) / a$ целое, т.к. $\operatorname{Tr}\left(a_{i} \bar{a}_{j}\right)$ делится на $a$. Задача 21. Если $I$ делится на $J$, то $I=J H \subset J$. Пусть $I \subset J$. По задаче 20 (которая применима, т.к. $J=(\alpha, \beta)$ по задаче 19) имеем $J \bar{J}=(a)$, т.е. $I \bar{J} \subset J \bar{J}=(a)$. Тогда все элементы $I \bar{J}$ делятся на $a$, откуда $H=I \bar{J} / a=\{x / a \mid x \in I \bar{J}\}$ - идеал. Получаем $J H=$ $(I \bar{J} / a) J=(I \bar{J} J) / a=(I(a)) / a=I$, т.е. $I$ делится на $J$. Упражнение 18. Доказано в решении задачи 20. Задача 22. По задаче 21 достаточно доказать, что $$ I \subset H \text { и } J \subset H \Leftrightarrow I+J \subset H $$ Из левого следует правое, поскольку $H$ замкнут относительно сложения; из правого левое - поскольку $I, J \subset I+J$. Упражнение 19. $(a) \overline{(a)}=(a \bar{a})=(N(a))=(|N(a)|)$ и $|N(a)| \geq 0$, т.е. $N((a))=|N(a)|$. Упражнение 20. Т.к. $N(I) N(J) \geq 0$, то достаточно показать, что $(N(I) N(J))=I J \overline{I J}$. Ho $(N(I) N(J))=(N(I))(N(J))=I \bar{I} J \bar{J}=I J \overline{I J}$. Упражнение 21. Если $J=I H$, то $N(J)=N(I) N(H)$ по упражнению 20 , откуда $N(I)$ делит $N(J)$. Упражнение 22. Если $I=(1)$, то $(N(I))=I \bar{I}=(1)$, т.е. $N(I)=1$. Если $N(I)=1$, то $1 \in I \bar{I} \subset I$, т.е. $I=(1)$. Упражнение 23. Идеал $I$ прост $\Leftrightarrow$ любой идеал, делящий $I$, совпадает либо с $I$, либо с (1) $\Leftrightarrow$ (по задаче 21 ) любой идеал, содержащий $I$, совпадает либо с $I$, либо с (1) $\Leftrightarrow$ идеал $I$ максимален. Упражнение 24. Пусть $p_{1}, p_{2}$ - различные простые идеалы. По упражнению 23 ни один из них не содержится в другом. Тогда $p_{1}+p_{2}$ - идеал, содержащий каждый из них и не совпадающий ни с одним из них. По упражнению 23 получаем $p_{1}+p_{2}=(1)$. Задача 23. Заметим, что если $N(I)$ - простое число, то $I$ - простой идеал (по упражнениям 21 и 22) и вычислим нормы: $N\left(p_{1}\right)=N\left(p_{2}\right)=3$, $N\left(p_{3}\right)=N\left(p_{4}\right)=5$. Задача 24. Используем индукцию по $N(I)$ и упражнение 20. Задача 25. По задаче 21 мы знаем, что $J H \subset I$ и хотим доказать что $H \subset I$. Пусть $h \in H$. Т.к. $I+J=(1)$, то найдутся $i \in I, J \in J$, что $i+j=1$. Тогда $j h \in J H \subset I$ и по определению идеала $i h \in I$. Значит, $h=j h+i h \in I$, что и требовалось. Задача 26. Пункт а) очевиден. б) $H I=H J \Rightarrow(N(H)) I=H \bar{H} I=H \bar{H} J=(N(H)) J$. Теперь применим пункт а) для $a=N(H)$. Задача 27. Предположим, что какой-то идеал имеет два различных разложения на простые идеалы. Если какой-то простой идеал встречается в обоих разложениях, то на него по предыдущей задаче можно сократить. Будем сокращать, пока не получим два разложения идеала $p_{1} \ldots p_{n}=I=q_{1} \ldots q_{n}$ на простые идеалы, такое что все $q_{i}$ отличны от $p_{1}$. Тогда $p_{1}$ делит $q_{1} \ldots q_{n-1} q_{n} ; p_{1}$ и $q_{n}$ взаимно просты по упражнению 24. По задаче $25 p_{1}$ делит $q_{1} \ldots q_{n-1}$. Используем индукцию и получаем противоречие. Задача 28. а) Пусть $I \subset \mathbb{Z}[\sqrt{-5}]$ - идеал, $x \in I$ - элемент с минимальной ненулевой нормой. Числа, кратные $x$, образуют на комплексной плоскости решетку из прямоугольников со сторонами 1 и $\sqrt{5}$. Если $I$ не содержит других чисел, то $I$ главный. Пусть $y \in I$, причем $y$ не делится на $x$. После параллельного переноса на число, кратное $x$, можно считать, что $y$ лежит в прямоугольнике с вершинами $0, x,(1+\sqrt{-5}) x, \sqrt{-5} x$. Мнимая часть $y / x$ лежит между 0 и $\sqrt{5}$. Если она меньше $\sqrt{3} / 2$ или больше $\sqrt{5}-\sqrt{3} / 2$, то расстояние от $y$ до одной из вершин прямоугольника меньше $|x|$, что невозможно. Поэтому $\frac{\sqrt{3}}{2} \leq \operatorname{Im} \frac{y}{x} \leq \sqrt{5}-\frac{\sqrt{3}}{2}$. Значит, $\sqrt{5}-\frac{\sqrt{3}}{2}<\sqrt{3} \leq 2 \operatorname{Im} \frac{y}{x} \leq 2 \sqrt{5}-\sqrt{3}<\sqrt{5}+\frac{\sqrt{3}}{2}$. Отсюда следует, что расстояние от $2 y$ до одного из чисел $\sqrt{-5} x,(1+\sqrt{-5}) x,(2+\sqrt{-5}) x$ меньше $|x|$, т.е. должно равняться нулю. Значит, $y$ равняется $\sqrt{-5} x / 2,(1+\sqrt{-5}) x / 2$, либо $(2+\sqrt{-5}) x / 2$. В первом и третьем случае $-5 x / 2 \in I$, откуда $x / 2 \in I$, что невозможно. То есть $y=(1+\sqrt{-5}) x / 2$ и $I=((1+\sqrt{-5}) a, 2 a)$, где $a=x / 2$. б) Взяв $x$ и $y$ аналогично пункту а) получим, что $y$ равняется $\sqrt{-6} x / 2$, $(1+\sqrt{-6}) x / 2$, либо $(2+\sqrt{-6}) x / 2$. Во втором случае $\sqrt{-6}(1+\sqrt{-6}) x / 2=$ $\sqrt{-6} x / 2-3 x \in I$, откуда $\sqrt{-6} x / 2 \in I$, откуда $x / 2=(1+\sqrt{-6}) x / 2-$ $\sqrt{-6} x / 2 \in I$, что невозможно. В первом и третьем случае получаем $I=$ $(\sqrt{-6} a, 2 a)$, где $a=x / 2$. Задача 29. Пусть $0 \neq a \in I$. Тогда $0 \neq a \bar{a} \in I \cap \mathbb{Z}$. Задача 30. Пусть $p$ - минимальное натуральное число в $I$, которое существует по предыдущей задаче. Если $p=a b$, где $a, b$ натуральные, то $(a)(b)=(a b)$ содержится в $I$, т.е. по задаче $21(a)(b)$ делится на $I$, но $(a)$ и (b) не делятся на $I$. Противоречие. Задача 31. Возьмем $p$ из предыдущей задачи. $(p) \subset I$, т.е. по задаче $21(p)$ делится на $I$, откуда $N((p))=p^{2}$ делится на $N(I)$. По упражнению $22 N(I) \neq 1$, откуда $N(I)$ равняется $p$ или $p^{2}$. Задача 32. Следует из решения предыдущей задачи. Задача 33. Если идеал $(p) \subset \mathbb{Z}[\omega]$ не прост, то $(p)=I J$, где $I, J$ отличны от (1). тогда $N(I) N(J)=N((p))=p^{2}$, т.е. $N(I)=p$, откуда $I \bar{I}=(N(I))=(p)$, что и требовалось. Задача 34. Предположим, что $0 \leq a1$ и $\operatorname{gcd}(p, q)=1$, и приведённый многочлен $P(x)=x^{n}+a_{n-1} x^{n-1}+\ldots+a_{1} x+a_{0}$ с целыми коэффициентами такой, что $P(a)=0$. Тогда $q^{n} P(a)=p^{n}+a_{n-1} p^{n-1} q+\ldots+a_{1} p q^{n-1}+a_{0} q^{n}=$ $p^{n}+q\left(a_{n-1} p^{n-1}+\ldots+a_{1} p q^{n-2}+a_{0} q^{n-1}\right)=0$, значит, $p^{n}$ делится на $q$. Но $\operatorname{gcd}(p, q)=1$, противоречие. Задача 36. Пусть $a, b \in \overline{\mathbb{Q}}$, тогда существуют приведённые многочлены $P(x), Q(x)$ с рациональными коэффициентами такие, что $P(a)=$ $Q(b)=0$. Пусть $P(x)=\left(x-\alpha_{1}\right)\left(x-\alpha_{2}\right) \ldots\left(x-\alpha_{n}\right)$ и $Q(x)=\left(x-\beta_{1}\right)(x-$ $\left.\beta_{2}\right) \ldots\left(x-\beta_{m}\right)$, где $\alpha_{1}, \alpha_{2} \ldots \alpha_{n}$ и $\beta_{1}, \beta_{2} \ldots \beta_{m}$ - корни $P(x)$ и $Q(x)$ соответственно, причем $a=\alpha_{1}$ и $b=\beta_{1}$. Тогда коэффициентами $P(x)$ и $Q(x)$ будут (с точностью до знака) элементарные симметрические многочлены от $\alpha_{1}, \alpha_{2} \ldots \alpha_{n}$ и $\beta_{1}, \beta_{2} \ldots \beta_{m}$ соответственно. Докажем, что $a+b \in \overline{\mathbb{Q}}$. Заметим, что $a+b$ является корнем многочлена $R(x)=\prod_{\substack{1 \leq i \leq n \\ 12$ уравнения 2 -ое, умноженное на старый (до последнего умножения) коэффициент $i$-ого уравнения при $\beta_{2}$. Покажем по индукции, что мы и дальше сможем по очереди исключать переменные. Предположим, что для натурального $1 \leq k \leq n$ - при $1 \leq i \leq k$, мы уже исключили $i$-ую переменную из уравнений с номером, большим $i$, и коэффициенты уравнений являются многочленами с целыми коэффициентами от $\alpha$, причем коэффициенты уравнений с номером $i>k$ при $\beta_{i}$ - степени $2^{k}$, степени остальных коэффициентов уравнений с номером $i>k$ строго меньше $2^{k}$, и все коэффициенты $i$-ых уравнений при $\beta_{i}-$ с точностью до знака приведенные многочлены: $$ \left(\begin{array}{cclccccc} P_{11}(\alpha) & P_{12}(\alpha) & \ldots & & & & & \\ 0 & P_{22}(\alpha) & \ldots & & & & & \\ \vdots & \vdots & \ddots & & & & & \\ 0 & 0 & \ldots & P_{k, k}(\alpha) & P_{k, k+1}(\alpha) & P_{k, k+2}(\alpha) & \ldots & P_{k, n}(\alpha) \\ 0 & 0 & \ldots & 0 & P_{k+1, k+1}(\alpha) & P_{k+1, k+2}(\alpha) & \ldots & P_{k+1, n}(\alpha) \\ 0 & 0 & \ldots & 0 & P_{k+2, k+1}(\alpha) & P_{k+2, k+2}(\alpha) & \ldots & P_{k+2, n}(\alpha) \\ \vdots & \vdots & \ldots & \vdots & \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \ldots & 0 & P_{n, k+1}(\alpha) & P_{n, k+2}(\alpha) & \ldots & P_{n, n}(\alpha) \end{array}\right) $$ где $\operatorname{deg}\left(P_{i i}\right)=2^{k}$ при $i>k$, и $\operatorname{deg}\left(P_{i j}\right)<2^{k}$ при $i, j>k$ и $i \neq j$, причем $P_{i i}$ - приведенный с точностью до знака. Рассмотрим уравнения с номерами $i>k$ от $\beta_{i}$ с $i>k$ : $$ \left(\begin{array}{cccc} P_{k+1, k+1}(\alpha) & P_{k+1, k+2}(\alpha) & \ldots & P_{k+1, n}(\alpha) \\ P_{k+2, k+1}(\alpha) & P_{k+2, k+2}(\alpha) & \ldots & P_{k+2, n}(\alpha) \\ \vdots & \vdots & \ddots & \vdots \\ P_{n, k+1}(\alpha) & P_{n, k+2}(\alpha) & \ldots & P_{n, n}(\alpha) \end{array}\right) $$ Для краткости обозначим $Q_{i j}=P_{k+i, k+j}(\alpha)$. По предположению $Q_{11} \neq$ 0 , поэтому мы можем умножить все уравнения после 1 -ого на $Q_{11}$, а потом вычесть из $i$-ого, $i>1$ уравнения 1 -ое, умноженное на старый (до последнего умножения) коэффициент $i$-ого уравнения при $\beta_{k+1}$ : $$ \left(\begin{array}{cccc} Q_{11} & Q_{12} & \cdots & Q_{1, n-k} \\ 0 & Q_{22} Q_{11}-Q_{21} Q_{12} & \cdots & Q_{2, n-k} Q_{11}-Q_{21} Q_{1, n-k} \\ \vdots & \vdots & \ddots & \vdots \\ 0 & Q_{n-k, 2} Q_{11}-Q_{n-k, 1} Q_{12} & \cdots & Q_{n-k, n-k} Q_{11}-Q_{n-k, 1} Q_{1, n-k} \end{array}\right) $$ Тогда $\operatorname{deg}\left(Q_{i i} Q_{11}-Q_{i, 1} Q_{1, i}\right)=2^{k+1}$, он приведенный с точностью до знака, и $\operatorname{deg}\left(Q_{i j} Q_{11}-Q_{i, 1} Q_{1, j}\right)<2^{k+1}$ при $i \neq j$. Мы исключили $\beta_{k+1}$ из уравнений с номером $i>k+1$ с сохранением условий на коэффициенты, тем самым переход индукции доказан (база $k=0$ ). Подставляя $k=n$, получим систему уравнений на $\beta_{1}, \ldots, \beta_{n}$, последнее уравнение которой имеет вид $P_{n, n}(\alpha) \beta_{n}=0$, где $P_{n, n}$ - приведенный многочлен с целыми коэффициентами. Значит, $P_{n, n}(\alpha) \neq 0$ и $\beta_{n} \neq 0$, противоречие. Значит, $\alpha$ - целое алгебраическое, ч.т.д. Задача 69. Пусть $\alpha_{1}, \ldots, \alpha_{n}$ - целый базис $I$, тогда если $J I=I$, то для каждого $i$ существуют такие $\beta_{i 1}, \ldots, \beta_{i n} \in J$, что $\alpha_{i}=\beta_{i 1} \alpha_{1}+$ $\ldots+\beta_{i n} \alpha_{n}$. Тем самым мы получаем систему линейных уравнений на $\alpha_{1}, \ldots, \alpha_{n}$ с коэффициентами $$ \left(\begin{array}{cccc} \beta_{11}-1 & \beta_{12} & \ldots & \beta_{1 n} \\ \beta_{21} & \beta_{22}-1 & \ldots & \beta_{2 n} \\ \vdots & \vdots & \ddots & \vdots \\ \beta_{n 1} & \beta_{n 2} & \ldots & \beta_{n n}-1 \end{array}\right) $$ Заметим, что ситуация аналогична с предыдущей задачей - целые числа заменены на идеал $J$, а $\alpha$ - на 1 . Аналогично с предыдущей задачей мы можем доказать, что 1 - корень приведенного многочленами с коэффициентами в $J$, из чего будет следовать, что $1 \in J$, поэтому $J=(1)$. Задача 70. Очевидно следует из конечности числа классов идеалов. Задача 71. По предыдущей задаче $(\alpha) I^{k}=(\beta) I^{m}$ для каких-то ненулевых $\alpha$ и $\beta$. Деля на $\beta$, получаем $(\alpha / \beta) I^{k}=I^{m}$, и т.к. $I^{m} \subseteq I^{k}$, то $(\alpha / \beta) I^{k} \subseteq I^{k}$. По задаче 68 получаем, что $\gamma=\alpha / \beta \in \widetilde{\mathbb{Z}}$, т.е. $(\gamma) I^{k}=$ $I^{m}=I^{k} I^{m-k}$. Для любого $\alpha \in I^{m-k}$ получаем $(\alpha) I^{k} \subseteq(\gamma) I^{k}$, значит $(\alpha / \gamma) I^{k} \subseteq I^{k}$, и по задаче $68 \alpha / \gamma \in \widetilde{\mathbb{Z}}$, т.е. все элементы $I^{m-k}$ делятся на $\gamma$. Поэтому $(1 / \gamma) I^{m-k}$ - идеал, и $I^{k}=(1 / \gamma) I^{m-k} I^{k}$. Значит, по задаче 69 мы получаем $(1 / \gamma) I^{m-k}=(1)$, поэтому $I^{m-k}=(\gamma)$. В частности, для $J=I^{m-k-1}-I J=(\gamma)$. Задача 72. Если $I$ делится на $J$, то $I=J H$ для какого-то идеала $H$, поэтому $I \subseteq J$. Обратно, если $I \subseteq J$, то возьмем по предыдущей задаче такой идеал $J^{\prime}$, что $J J^{\prime}=(\alpha)$ для ненулевого $\alpha$. Тогда $I J^{\prime} \subseteq J J^{\prime}=(\alpha)$, поэтому $H=(1 / \alpha) I J^{\prime}$ - идеал, и $I=(1 / \alpha) J J^{\prime} I=J H$. Значит, $I$ делится на $J$. Задача 73. Для ненулевого идеала $I$ обозначим $N(I)=|\widetilde{\mathbb{Z}} / I|$ - количество элементов в $\widetilde{\mathbb{Z}} / I$. Если $I \subset J$, то $N(I)>N(J)$, т.к. классы эквивалентности $\widetilde{\mathbb{Z}} / J$ являются объединением каких-то классов эквивалентности $\widetilde{\mathbb{Z}} / I$, причем класс $J$ содержит больше одного класса $\widetilde{\mathbb{Z}} / I$, т.к. $I \neq J$. Для ненулевого идеала $I$ найдем какой-то его простой делитель, продолжая по индукции цепочку вложенных собственных идеалов $I \subset I_{1} \subset$ $I_{2} \subset \ldots-$ т.к. $N\left(I_{k}\right)$ уменьшается, то в какой-то момент мы не сможем её продолжить, тогда последний член в цепочке будет простым идеалом $P_{1}$ (т.к. он не содержится в других собственных идеалах). По предыдущей задаче $I=P_{1} H$ для какого-то идеала $H$, причем $N(H)1$ ), мы получаем разложение $I=P_{1} P_{2} \ldots P_{n}$ на простые идеалы. Задачи 74 - 78. Аналогично задачам $25-27$. # Generalizations of the fundamental theorem of arithmetic Vera Bulankina, Ivan Frolov, Timofei Zaitsev,
Aleksei Petukhov, Ruslan Salimov* ## Introduction The goal of this project is to generalize the fundamental theorem of arithmetic from integers to some more "advanced" objects. For solving the problems of the project, it is possible to team up with other participants. The fundamental theorem of arithmetic (Theorem 1 below, see also Theorems 4, 5, 7) is useful for finding integer solutions of various polynomial equations, i.e. Diophantine equations. For example they are useful in some proofs of Fermat's theorem on sum of two squares (Theorem 2) and Fermat's Last Theorem (Theorem 3) for $n=3$. You can try to prove them right now, but most probably it is better to return to them after the relevant sections of the project. Also you will see that the uniqueness of the prime factorization does not always hold for analogs of integers. In the second part of the project we consider the version of the fundamental theorem of arithmetic which works for numbers of the form $a+b \sqrt{d}$, where $d$ is a fixed integer and $a, b$ are any integer variables (to do this, we need a concept of ideal). This technique will be applied to Diophantine equations. In the third part we will discuss a more general statement on arbitrary algebraic numbers. Finally, there is an additional list of problems in which we will try to apply the general theory to Fermat's Last Theorem. We will discuss the so-called first case of the Fermat's Last Theorem for the regular prime numbers. Using similar ideas one can prove the second case of the Fermat's Last Theorem for any $n$, which is divisible by a regular prime number $p$. All prime number up to 37 are regular, see books of M. Postinkov or J. Milne or Wikipedia. We would like to mention that Andrew Wiles general proof of the Fermat's Last Theorem is based on essentially different methods and ideas. We have have used books of K. Ireland and M. Rosen, J. Milne, M. Postnikov, L. Washington, notes of K. Conrad and Wikipedia to prepare the project. We also added a few references [SS, Go, ZSS] that may be of interest to the reader who wants to learn more about this subject. Theorem 1. The fundamental theorem of arithmetic. Every integer $n>1$ is a product $n=p_{1} \cdot \ldots \cdot p_{k}$ of prime numbers $p_{1}, \ldots, p_{k}$. Moreover, this presentation is unique, up to an order of the factors. Theorem 2. Fermat's theorem on sum of two squares. An integer $n>0$ can be expressed as a sum of two squares if and only if all prime numbers of the form $4 k+3$ participate the prime decomposition of $n$ even number of times. Theorem 3. Fermat's Last Theorem. The equation $x^{n}+y^{n}=z^{n}$ has no solution in positive integers $x, y$ and $z$ for $n>2$.[^1] ## 1 Gaussian integers In this part of the project we want to discuss Gaussian integers. This is a generalization of integers which uses $\sqrt{-1}$. If you can't solve some of the problems from this section, then try to solve problems from other sections of the project and come back here later. Definition. Gaussian integers is the set of complex numbers of the form $a+b i$, where $a, b \in \mathbb{Z}$, $i=\sqrt{-1}$. We denote the set of Gaussian integers by $\mathbb{Z}[i]$. Problem 1. Prove that the sum of any two Gaussian integers (and the product of any two Gaussian integers) is a Gaussian integer. To proceed with the unique factorization for Gaussian integers we need a more exact formulation of the fundamental theorem of arithmetic. Strictly speaking, a straightforward analog of the statement of Theorem 1 is not literally correct even for integers: $2 \cdot 3=6=(-2) \cdot(-3)$. For the new formulation we need few more definitions. Definition. We say that an element $z \in \mathbb{Z}[i]$ is invertible if there is $b \in \mathbb{Z}[i]$ such that $a b=1$. The invertible elements of $\mathbb{Z}$ are defined in the same way. Exercise 1. Prove that the invertible elements in integers are 1 and -1 . Prove that the invertible elements in Gaussian integers are $1,-1$ and also $i,-i$. Definition. A Gaussian integer is called prime, if for any it's factorization into two factors strictly one of these factors is an invertible element. Definition. Two factorizations into prime numbers are the same if they have the same number of factors and if we can change the order of factors in such a way that the ratio of the corresponding prime factors is an invertible element. For example $7 \cdot 3,(-3) \cdot(-7)$ and $(-7 i) \cdot(3 i)$ are the same factorizations of 21 in Gaussian integers. Our immediate goal is to prove and to discuss the following theorem. Theorem 4. The fundamental theorem of arithmetic for the Gaussian integers. Any two factorizations of a Gaussian integer into prime numbers are the same. Problem 2* Define a division-with-remainder for $\mathbb{Z}[i]$. Use it to prove Theorem 4. Hint: use magnitude of complex numbers and graphical (Gaussian) interpretation of complex numbers. Problem 3. Find all integer solutions of the equation $x^{2}+1=y^{n}$. Problem 4. a) Let $p \in \mathbb{Z}$ be prime. Prove that $p$ is a prime Gaussian integer if and only if $p+1$ is divisible by 4 . b) If $n, m \in \mathbb{Z}$ can be expressed as a sum of two squares, then $m n$ can be expressed as a sum of two squares. c) Prove Fermat's theorem on sum of two squares (Theorem 2). Problem 5. Let $n$ be an integer. Count the number of ways in which $n$ can be decomposed into the sum of squares? Hint: try to use the prime factorization of $n$. ## Eisenstein integers The goal of this section is to solve of the following problem. Problem 6* Prove Theorem 3 for $n=3$ using the following formula $$ x^{3}+y^{3}=(x+y)\left(x+\frac{-1+\sqrt{-3}}{2} y\right)\left(x+\frac{-1-\sqrt{-3}}{2} y\right) $$ Let's introduce several definitions and consider several ancillary problems to achieve this goal. Denote by $\xi$ a cubic root of unity such that $\xi \neq 1$. Exercise 2. Prove that $\xi=\frac{-1 \pm \sqrt{-3}}{2}$. Put $\mathbb{Z}[\xi]:=\{a+b \xi: a, b \in \mathbb{Z}\}$. Such numbers are called Eisenstein integers. Definition. An Eisenstein integer $\alpha \in \mathbb{Z}[\xi]$ is divisible by $\beta \in \mathbb{Z}[\xi]$ if and only if there is $\gamma \in \mathbb{Z}[\xi]$ such that $\alpha=\beta \gamma$. The invertible elements of $\mathbb{Z}[\xi]$ are defined similarly to the invertible elements of $\mathbb{Z}$ and of $\mathbb{Z}[i]$. Definition. An Eisenstein integer $\alpha \in \mathbb{Z}[\xi]$ is composite if $\alpha=\beta \gamma$ where $\beta$ and $\gamma \in \mathbb{Z}[\xi]$ aren't invertible. An nonzero Eisenstein integer $\alpha \in \mathbb{Z}[\xi]$ is prime, if $\alpha$ is neither composite nor invertible. Problem 7. Find out all invertible elements of $\mathbb{Z}[\xi]$. Problem 8. Define a division-with-remainder procedure for $\mathbb{Z}[\xi]$. Use it to formulate and to prove the fundamental theorem of arithmetic for $\mathbb{Z}[\xi]$. ## Quadratic fields In this section we consider quadratic fields. They generalise Gaussian integers and Eisenstein integers. Definition. For complex numbers $a_{1}, \ldots, a_{n}$ let $\mathbb{Z}\left[a_{1}, \ldots, a_{n}\right]$ denote the smallest subset of $\mathbb{C}$ containing $\mathbb{Z}$ and $a_{1}, \ldots, a_{n}$ which is closed under addition, subtraction and multiplication. Define $\mathbb{Q}\left[a_{1}, \ldots, a_{n}\right]$ similarly (replace $\mathbb{Z}$ by $\mathbb{Q}$ ). The sets $\mathbb{Z}\left[a_{1}, \ldots, a_{n}\right]$ play an analogous role in $\mathbb{Q}\left[a_{1}, \ldots, a_{n}\right]$ as the ordinary integers do in $\mathbb{Q}$. (Their elements can be added and multiplied). Fix a square-free integer $d \neq 1$. Remark. There are two main examples for $d: d=-3$ and $d=2$. It can be useful to first check all problems of this section for these values of $d$, and then proceed to the general case. Exercise 3. a) Prove that $$ \mathbb{Q}[\sqrt{d}]=\{x+y \sqrt{d}: x, y \in \mathbb{Q}\} \text { and } \mathbb{Z}[\sqrt{d}]=\{x+y \sqrt{d}: x, y \in \mathbb{Z}\} $$ b) Prove that if $a, b \in \mathbb{Q}[\sqrt{d}]$ and $b \neq 0$, then $\frac{a}{b} \in \mathbb{Q}[\sqrt{d}]$. Definition. Integers in $\mathbb{Q}[\sqrt{d}]$ are numbers $\alpha \in \mathbb{Q}[\sqrt{d}]$, satisfying $\alpha^{2}+p \alpha+q=0$ with $p, q \in \mathbb{Z}$. Exercise 4. Is $\xi$ an integer in $\mathbb{Q}[\sqrt{-3}]$ ? Is $\frac{1+i}{2}$ an integer in $\mathbb{Q}[i]$ ? Set $\omega=\sqrt{d}$ when $d \equiv 2,3 \bmod 4$, and $\omega=\frac{\sqrt{d}+1}{2}$ when $d \equiv 1 \bmod 4$. The unique factorization into prime numbers can fail for $\mathbb{Z}[\omega]$, but there are several valid modifications of it. Further we will discuss one of them. For any $\alpha=a+b \sqrt{d}$, where $a, b \in \mathbb{Q}$, we define its conjugate $\bar{\alpha}=a-b \sqrt{d}$, its norm $N(\alpha)=\alpha \bar{\alpha}$ and its trace $\operatorname{Tr}(\alpha)=\alpha+\bar{\alpha}$. Exercise 5. Prove that $\overline{a+b}=\bar{a}+\bar{b}$ and $\overline{a b}=\bar{a} \bar{b}$. Exercise 6. a) Prove that $\alpha \in \mathbb{Q}[\sqrt{d}]$ is a root of the polynomial $x^{2}-\operatorname{Tr}(\alpha) x+N(\alpha)$. b) Prove that $\alpha \in \mathbb{Q}[\sqrt{d}]$ is an integer in $\mathbb{Q}[\sqrt{d}]$ if and only if $N(\alpha) \in \mathbb{Z}$ and $\operatorname{Tr}(\alpha) \in \mathbb{Z}$. Problem 9. Prove that the set of integers in $\mathbb{Q}[\sqrt{d}]$ coincides with $\mathbb{Z}[\omega]$. The invertible elements of $\mathbb{Z}[\omega]$ are defined similarly to the invertible elements of $\mathbb{Z}$ and $\mathbb{Z}[i]$. Exercise 7. Prove that $\gamma \in \mathbb{Z}[\omega]$ satisfies $N(\gamma)= \pm 1$ if and only if $\gamma$ is invertible. Definition. A number $\alpha \in \mathbb{Z}[\omega]$ is divisible by $\beta \in \mathbb{Z}[\omega]$ if there exists $\gamma \in \mathbb{Z}[\omega]$ such that $\alpha=\beta \gamma$. Definition. A nonzero number $\alpha \in \mathbb{Z}[\omega]$ is composite if $\alpha=\beta \gamma$, where $\beta$ and $\gamma \in \mathbb{Z}[\omega]$ are not invertible. A number $\alpha \in \mathbb{Z}[\omega]$ is prime if $\alpha$ is neither composite nor invertible. Exercise 8. For $\gamma \in \mathbb{Z}[\omega]$ prove that if $|N(\gamma)| \in \mathbb{Z}$ is prime then $\gamma$ is prime in $\mathbb{Z}[w]$. Show that the converse is false in $\mathbb{Z}[\sqrt{3}]$. Exercise 9. Prove that if $\gamma \in \mathbb{Z}[\omega]$ is not invertible, then $\gamma$ is a product of primes in $\mathbb{Z}[\omega]$. Problem 10. Prove that all factors in the following decomposition of 15 in $\mathbb{Z}[\sqrt{-14}]$ are prime. $$ 15=3 \cdot 5=(1+\sqrt{-14})(1-\sqrt{-14}) $$ Problem 11. Define a version of Euclidean algorithm for a) $\mathbb{Z}[\sqrt{-2}]$; b) $\mathbb{Z}\left[\frac{1+\sqrt{-7}}{2}\right]$. Use it to prove the unique factorization in $\mathbb{Z}[\sqrt{-2}]$ and $\mathbb{Z}\left[\frac{1+\sqrt{-7}}{2}\right]$. Problem 12. Find all invertible elements of $$ \text { a) } \mathbb{Z}[\sqrt{-1}] \text {, b) } \mathbb{Z}[\sqrt{-d}] \text { with } d \geq 2, \text { c) } \mathbb{Z}\left[\frac{1+\sqrt{-3}}{2}\right] \text {, d*) } \mathbb{Z}[\sqrt{2}] $$ Problem 13. Does unique factorization hold in a) $\mathbb{Z}[\sqrt{2}]$, b) $\mathbb{Z}[\sqrt{-3}]$, c) $\mathbb{Z}[\sqrt{3}]$, d) $\mathbb{Z}[\sqrt{-5}]$, e) e) $\mathbb{Z}[\sqrt{5}]$, f) $\mathbb{Z}[\sqrt{10}]$, g) $\mathbb{Z}\left[\frac{1+\sqrt{5}}{2}\right]$ h) $\mathbb{Z}\left[\frac{1+\sqrt{-7}}{2}\right]$, i) $\left.\mathbb{Z}\left[\frac{1+\sqrt{-11}}{2}\right], j^{*}\right) \mathbb{Z}\left[\frac{1+\sqrt{-19}}{2}\right]$ ? Problem 14. For which complex numbers $\xi$ the sum and the product of any numbers of the form $a+b \xi$ with $a, b \in \mathbb{Z}$ have the same form? Problem 15* Find all positive integer solutions of a) $3^{n}=k^{2}+2$ b) $2^{n}=k^{2}+7$. ## Ideals In the following sections we give an approach to the following problems. Problem 16. Find all integer solutions of a) $x^{2}+5=y^{3}$, b) $x^{2}+2 x+7=y^{3}$ c) $5 x^{2}+1=y^{3}$, d) $\left.6 x^{2}-12 x+7=y^{3}, \quad \mathrm{e}^{*}\right) x^{2}-6=y^{3}$. Definition. A nonempty subset $I \in \mathbb{Z}$ is called an ideal if it is closed under addition, subtraction and multiplication by integers: $$ a, b \in I \Longrightarrow a \pm b \in I, \quad a \in I, b \in \mathbb{Z} \Longrightarrow a b \in I $$ Remark. The notion of ideal was invented by Richard Dedekind as a replacement for the ideal numbers of Ernst Kummer (these numbers turn out to be useful to solve some cases Fermat's last theorem). One can check whether or not a given number is divisible by an ideal number even if this ideal number is not well defined. A similar idea is behind the definition of Dedekind cut of rational numbers. Exercise 10. a) Prove that any ideal contains 0. b) Prove that $a \in I$ implies $-a \in I$. c) Prove that the set of all even integers form an ideal in $\mathbb{Z}$. d) Prove that the set $\{2018 m: m \in \mathbb{Z}\}$ is an ideal in $\mathbb{Z}$. Exercise 11. Prove that the intersection of several ideals is an ideal. We denote by $\left(a_{1}, \ldots, a_{n}\right)$ the smallest ideal (intersection of all ideals) containing $a_{1}, \ldots, a_{n} \in \mathbb{Z}$. Problem 17. a) Let $a, b$ be relatively prime numbers. Prove that $(a, b)=(1)=\mathbb{Z}$. b) Prove that $(a, b)=(g)$, where $a, b \in \mathbb{Z}$ and $g$ is the greatest common divisor of $a, b$. c) Prove that every ideal $I \neq\{0\}$ in $\mathbb{Z}$ is $(g)$ for some $g \in \mathbb{Z}$. As in the previous section we set $\omega=\sqrt{d}$ when $d \equiv 2,3 \bmod 4$, and $\omega=\frac{\sqrt{d}+1}{2}$ when $d \equiv 1 \bmod 4$. An ideal in $\mathbb{Z}[\omega]$ is defined similarly to $\mathbb{Z}(\mathbb{Z}$ is replaced by $\mathbb{Z}[\omega])$. Similarly, numbers $\alpha_{1}, \ldots, \alpha_{n} \in \mathbb{Z}[\omega]$ define an ideal $\left(\alpha_{1}, \ldots, \alpha_{n}\right)$ in $\mathbb{Z}[\omega]$. In particular, every element $\alpha \in \mathbb{Z}[\omega]$ defines the ideal $(\alpha)$. Exercise 12. Prove that for nonzero $\alpha$ and $\beta$ in $\mathbb{Z}[w](\alpha)=(\beta)$ if and only if $\alpha / \beta, \beta / \alpha \in \mathbb{Z}[\omega]$ (i.e. $\alpha / \beta$ is invertible in $\mathbb{Z}[\omega]$ ). Exercise 13. Let $a, x, y \in \mathbb{Z}$. Prove that $x+y \omega \in(a)$ if and only if $a \mid x$ and $a \mid y$. Definition. An ideal of the form $(\alpha)$ for some $\alpha \in \mathbb{Z}[\omega]$ is called principal. As we have seen in Problem 17, all ideals in $\mathbb{Z}$ are principal. Problem 18. Show that $(2, \sqrt{-14})$ is not a principal ideal in $\mathbb{Z}[\sqrt{-14}]$. Problem 19. Prove that, for each ideal $I$ in $\mathbb{Z}[\omega]$, there exist $\alpha, \beta \in \mathbb{Z}[\omega]$ such that $$ I=\{x \alpha+y \beta: \quad x, y \in \mathbb{Z}\} $$ Definition. For two ideals $I, J \in \mathbb{Z}[\omega]$ we set $$ \begin{gathered} I+J:=\left\{i_{1}+i_{2}: i_{1} \in I_{1}, i_{2} \in J\right\}, \quad \bar{I}:=\{\bar{i}: i \in I\} \\ I J:=\left\{i_{1} j_{1}+\cdots+i_{k} j_{k}: i_{1}, \ldots, i_{k} \in I, j_{1}, \ldots, j_{k} \in J\right\} \end{gathered} $$ Exercise 14. In $\mathbb{Z}[\sqrt{-14}]$ compute the product of ideals $I=(5+\sqrt{-14}, 2+\sqrt{-14})$ and $J=$ $(4+\sqrt{-14}, 2-\sqrt{-14})$. Exercise 15. Show that $$ (3)=p_{1} p_{2}, \quad(5)=p_{3} p_{4}, \quad(1+\sqrt{-14})=p_{1} p_{3}, \quad(1-\sqrt{-14})=p_{2} p_{4} $$ where $$ p_{1}=(3,1+\sqrt{-14}), p_{2}=(3,1-\sqrt{-14}), p_{3}=(5,1+\sqrt{-14}), p_{4}=(5,1-\sqrt{-14}) $$ Exercise 16. Compute (20)(18), (20) $+(18),(20) \cap(18)$ in $\mathbb{Z}[\omega]$ for all $d$. Exercise 17. Prove that $I+J, \bar{I}, I J$ are ideals in $\mathbb{Z}[\omega]$ for all ideals $I, J \subset \mathbb{Z}[\omega]$. ## 2 Unique factorization: quadratic fields In Problem 10 we saw that the straightforward version of the unique factorization fails for $\mathbb{Z}[\sqrt{-14}]$ : $$ 3 \cdot 5=(1+\sqrt{-14})(1-\sqrt{-14}) $$ On the other hand, by Exercise 15, $$ (3)=p_{1} p_{2}, \quad(5)=p_{3} p_{4}, \quad(1+\sqrt{-14})=p_{1} p_{3}, \quad(1-\sqrt{-14})=p_{2} p_{4} $$ where $$ p_{1}=(3,1+\sqrt{-14}), p_{2}=(3,1-\sqrt{-14}), p_{3}=(5,1+\sqrt{-14}), p_{4}=(5,1-\sqrt{-14}) $$ i.e. $(15)=p_{1} p_{2} p_{3} p_{4}$. The ideals $p_{1}, p_{2}, p_{3}, p_{4}$ are a replacement of prime numbers, see Problem 23 , and the factorization $(15)=p_{1} p_{2} p_{3} p_{4}$ is unique up to the order of the factors, as the following theorem shows. Theorem 5. Fundamental theorem of arithmetic for quadratic fields. For every ideal $I \subset \mathbb{Z}[\omega]$ that is not (0) or (1) there exists a factorization of $I$ as a product of prime ideals into prime ideals $$ I=p_{1} \cdots p_{s} \subset \mathbb{Z}[\omega] $$ This factorization is unique up to an order of the factors. Theorem 5 has two parts: existence of the factorization and its uniqueness. The former is proved in Problem 24 and the latter in Problem 27. As an example, this theorem can be used to solve Problem 16. See also the section on Fermat's Last Theorem. Corollary. For every $m \in \mathbb{Z}[\omega]$ that is not 0 or invertible there exists a factorization of the ideal $(m)$ into a product of prime ideals $p_{1}, \ldots, p_{s} \subset \mathbb{Z}[\omega]$, unique up to the order of the factors. The proof is divided into the following problems. Problem 20. For every $a_{1}, \ldots, a_{n} \in \mathbb{Z}[\omega]$ prove that $$ \left(a_{1}, \ldots, a_{n}\right)\left(\bar{a}_{1}, \ldots, \bar{a}_{n}\right)=\left(N\left(a_{i}\right), \operatorname{Tr}\left(a_{i} \overline{a_{j}}\right)\right)_{1 \leq i, j \leq n} $$ Hint: consider the case of an ideal generated by two elements. Definition. We say that an ideal $I$ is divisible by an ideal $J$ if there exists an ideal $H$ in $\mathbb{Z}[\omega]$ such that $I=J H$. Problem 21. For any two ideals $I, J$ in $\mathbb{Z}[\omega]$ prove that $I$ is divisible by $J$ if and only if $I$ is contained in $J$. Hint: use the previous problem. Exercise 18. Use Problem 20 to prove that, for every nonzero ideal $I \subset \mathbb{Z}[\omega]$, there exists a positive integer $N(I)$ such that $I \bar{I}=(N(I))$. Problem 22. Prove that an ideal $H$ divides $I$ and $J$ if and only if it divides $I+J$. Exercise 19. Prove that $N((a))=|N(a)|$ for every nonzero $a \in \mathbb{Z}[\omega]$. Exercise 20. Prove that $N(I) N(J)=N(I J)$ for all nonzero ideals $I, J$ in $\mathbb{Z}[w]$. Exercise 21. Prove that if an ideal $I$ divides an ideal $J$, than $N(I)$ divides $N(J)$. Exercise 22. Prove that $N(I)=1$ if and only if $I=(1)$. An ideal $I$ in $\mathbb{Z}[\omega]$ is called prime if it is not (1) and it is divisible by exactly two ideals: itself and (1). Exercise 23. Prove that an ideal $I$ is prime if and only if it is maximal, i.e. the only bigger ideal is (1). Two ideals $I, J$ are called relatively prime if $I+J=(1)$. Exercise 24. Prove that any two distinct prime ideals are relatively prime. Problem 23. Show that the following ideals are prime in $\mathbb{Z}[\sqrt{-14}]$ : $$ p_{1}=(3,1+\sqrt{-14}), p_{2}=(3,1-\sqrt{-14}), p_{3}=(5,1+\sqrt{-14}), p_{4}=(5,1-\sqrt{-14}) $$ Problem 24. Prove that every nonzero ideal $I$ in $\mathbb{Z}[\omega]$ besides (1) equals the product of several prime ideals. Problem 25. Prove that if ideals $I, J$ in $\mathbb{Z}[\omega]$ are relatively prime and $I$ divides $H J$ for an ideal $H$, then $I$ divides $H$. Problem 26. a) Suppose that $a \in \mathbb{Z}[\omega] \backslash\{0\}$ and $(a) I=(a) J$ for ideals $I, J \subset \mathbb{Z}[\omega]$. Prove that $I=J$. b) Suppose that ideals $I, H, J$ satisfy $H \neq(0)$ and $H I=H J$. Prove that $I=J$. Problem 27. Prove that the factorization in Problem 24 is unique up to the order of the factors. ## Prime ideals and prime numbers Problem 28. a) Prove, that every ideal in $\mathbb{Z}[\sqrt{-5}]$ either is principal or equals $((1+\sqrt{-5}) a, 2 a)$ for some $a \in \mathbb{Q}[\sqrt{-5}]$. b) Prove that every ideal in $\mathbb{Z}[\sqrt{-6}]$ either is principal or equals $(\sqrt{-6} a, 2 a)$ for some $a \in \mathbb{Q}[\sqrt{-6}]$. Problem 29. Prove that every nonzero ideal in $\mathbb{Z}[\omega]$ contains a positive integer. Problem 30. Prove that every prime ideal $I$ in $\mathbb{Z}[\omega]$ contains a unique prime number $p \in \mathbb{Z}, p>0$. Problem 31. Suppose $I$ is a prime ideal. Prove that $N(I)$ equals either $p$ or $p^{2}$ for some prime number $p \in \mathbb{Z}$. Problem 32. Prove that prime numbers $p$ in two previous problems coincide. Problem 33. Prove that for any prime number $p \in \mathbb{Z}$ the ideal $(p) \subset \mathbb{Z}[\omega]$ either is prime or equals to a product of two (not nessesarily distinct) conjugate prime ideals. Problem 34. Let $P_{\omega}(x)$ be monic quadratic polynomial with integer coefficients such that $P_{\omega}(\omega)=0$. In assumptions of previous problem prove that former case takes place if and only if the equation $P_{\omega}(x)=0$ do not have solutions modulo $p$. ## Algebraic numbers In this section we wish to discuss algebraic numbers together with techniques and problems arising from this notion. The content of this section is present (as a general rule) in number theory courses of university level, but we believe that this content can be managed by advanced high school students as well. Definition. A complex number $\alpha \in \mathbb{C}$ is called algebraic if it is a root of a nonzero polynomial with rational coefficients. An algebraic number $\alpha \in \mathbb{C}$ is an algebraic integer if it is a root of a monic polynomial with integer coefficients. We denote the set of algebraic numbers by $\overline{\mathbb{Q}}$. We denote the set of algebraic integers by $\overline{\mathbb{Z}}$. Problem 35. Pick $a \in \mathbb{Q}$. Show that if $a \in \overline{\mathbb{Z}}$ then $a \in \mathbb{Z}$. Problem 36* Pick $\alpha, \beta \in \overline{\mathbb{Q}}$. Show that $\alpha \pm \beta \in \overline{\mathbb{Q}}, \alpha \beta \in \overline{\mathbb{Q}}, \alpha / \beta \in \overline{\mathbb{Q}}$ (in the latter case we assume that $b \neq 0)$. Hint: use Vieta's formulas. Problem 37* Let $\beta$ be a root of $\alpha_{n} x^{n}+\cdots+\alpha_{0}$ where $\alpha_{0}, \ldots, \alpha_{n} \in \overline{\mathbb{Q}}$. Show that $b \in \overline{\mathbb{Q}}$. Problem 38. a) Define a division-with-remainder procedure for the set of polynomials (in one variable) with coefficients in real numbers, complex numbers and rational numbers. b) Prove that the unique factorization property holds for the set of polynomials (in one variable) with rational coefficients. Problem 39. Gauss's lemma for polynomials Let $c_{g}$ be the greatest common divisor of the coefficients of $g(x) \in \mathbb{Z}[x]$. Show that, for all nonzero $g_{1}(x), g_{2}(x) \in \mathbb{Z}[x]$, we have $c_{g_{1} g_{2}}=c_{g_{1}} c_{g_{2}}$. Problem 40. Can the constructions and the statements of Problem 38 be applied to the set of polynomials with integer coefficients? What about the set of polynomials in two variables with complex coefficients? Problem 41. Pick an algebraic number $\alpha$. Let $P_{\alpha}(x)$ be a monic irreducible polynomial of the least degree such that $P_{\alpha}(\alpha)=0$. Prove that if $Q(\alpha)=0$ for a polynomial $Q(x)$ then $P_{\alpha}$ divides $Q$. Problem 42* Pick $\alpha, \beta \in \overline{\mathbb{Z}}$. Show that $\alpha \pm \beta \in \overline{\mathbb{Z}}, \alpha \beta \in \overline{\mathbb{Z}}$. Hint: use Vieta's formulas. Problem 43* Let $\alpha_{1}, \ldots, \alpha_{n}$ be algebraic numbers. Show that if $\alpha, \beta \in \mathbb{Q}\left[\alpha_{1}, \ldots, \alpha_{n}\right]$ and $\beta \neq 0$ then $\frac{\alpha}{\beta} \in \mathbb{Q}\left[\alpha_{1}, \ldots, \alpha_{n}\right]$. Problem 44* Pick an algebraic integer $\alpha$. Let $Q(x)$ be a monic irreducible polynomial with rational coefficients of minimal degree such that $Q(\alpha)=0$. Prove that the coeffecients of $Q(x)$ are integers. Problem 45* Let $f(x)$ be a monic polynomial with integer coefficients such that the absolute values of all roots of $f(x)$ in $\mathbb{C}$ are 1 . Show that all roots of $f(x)$ are roots of unity. ## 3 Ideal classes Definition. Let $\alpha_{1}, \ldots, \alpha_{k}$ be algebraic numbers. Set $\widetilde{\mathbb{Q}}:=\mathbb{Q}\left[\alpha_{1}, \ldots, \alpha_{k}\right]$ and $\widetilde{\mathbb{Z}}:=\overline{\mathbb{Z}} \cap \widetilde{\mathbb{Q}}$. Note that subsets of $\mathbb{C}$ which are closed under addition, subtraction and multiplication are called rings. Definition. A nonempty subset $I$ of $\widetilde{\mathbb{Z}}$ is called ideal if it is closed under addition, subtraction and multiplication by the elements of $\widetilde{\mathbb{Z}}$. Definition. We say that two nonzero ideals $I, J \subseteq \widetilde{\mathbb{Z}}$ are equivalent $(I \sim J)$ if there exist nonzero $\alpha, \beta \in \widetilde{\mathbb{Z}}$ such that $(\alpha) I=(\beta) J$. Problem 46. Check that $\sim$ is an equivalence relation. Definition. The equivalence classes of ideals are called ideal classes of $\widetilde{\mathbb{Z}}$. Problem 47. Show that the number of ideal classes in $\widetilde{\mathbb{Z}}$ equals 1 if and only if all ideals in $\widetilde{\mathbb{Z}}$ are principal. Problem 48. Let $I_{1}, I_{2}, J_{1}, J_{2}$ be nonzero ideals in $\widetilde{\mathbb{Z}}$ such that $I_{1} \sim I_{2}$ and $J_{1} \sim J_{2}$. Show that $I_{1} J_{1} \sim I_{2} J_{2}$. Problem 49. Describe the ideal classes of $\mathbb{Z}[\sqrt{-5}]$ and $\mathbb{Z}[\sqrt{-6}]$. Can you say anything about the multiplication of these classes? Problem 50. Let $I$ be a nonzero ideal in $\widetilde{\mathbb{Z}}$ and pick $\alpha \in \widetilde{\mathbb{Z}}$. If $I \subseteq(\alpha)$ then the set $(1 / \alpha) I$ is an ideal of $\widetilde{\mathbb{Z}}$. In the rest of this section we will discuss one the most fundamental statements of Algebraic Number Theory. This statement will play a key role in our proof of a version of the fundamental theorem of arithmetic. The latter proof will be discussed in the subsequent section under the assumption that Theorem 6 is already proven. Theorem 6. The number of ideal classes in $\widetilde{\mathbb{Z}}$ is finite. Definition. Consider $x_{1}, \ldots, x_{n} \in \widetilde{\mathbb{Q}}$. We say that $\left\{x_{1}, \ldots, x_{n}\right\}$ is a $\mathbb{Q}$-basis of $\widetilde{\mathbb{Q}}$ if every element $\alpha \in \widetilde{\mathbb{Q}}$ can be expressed uniqely in the form $$ m_{1} x_{1}+\cdots+m_{n} x_{n} $$ where $m_{1}, \ldots, m_{n} \in \mathbb{Q}$. Problem 51. Pick an algebraic number $\alpha$. Show that $\mathbb{Q}[\alpha]$ has a finite $\mathbb{Q}$-basis. Problem 52. Show that $\widetilde{\mathbb{Q}}$ has a finite $\mathbb{Q}$-basis. Problem 53. Pick an algebraic number $\alpha$. Show that there exists a nonzero $n \in \mathbb{Z}$ such that $n \alpha$ is an algebraic integer. Problem 54. Let $I \subset \widetilde{\mathbb{Z}}$ be a nonzero ideal. Show that there exists a finite set $\alpha_{1}, \ldots, \alpha_{N} \in I$ such that every $\alpha \in I$ equals $m_{1} \alpha_{1}+m_{2} \alpha_{2}+\ldots+m_{N} \alpha_{N}$ where $m_{1}, \ldots, m_{N} \in \mathbb{Z}$ (such $m_{1}, \ldots, m_{N}$ need not be unique). Hint: show that, for a fixed $\mathbb{Q}$-basis of $\widetilde{\mathbb{Q}}$, the coefficients $\alpha \in I$ can't be too small. Problem 55. Prove that there exists $\alpha_{1}, \ldots, \alpha_{n} \in I$ as in Problem 54 such that the respective representation is unique. Hint: use induction on the size of basis. Definition. We say that $\left\{\alpha_{1}, \ldots, \alpha_{n}\right\} \subset I$ is an integer basis of $I$ if they satisfy the conditions of Problem 55. Problem 56. Prove that an integer $\mathbb{Q}$-basis of $I$ is a $\mathbb{Q}$-basis of $\widetilde{\mathbb{Q}}$. Definition. Denote by $\widetilde{\mathbb{Z}} / I$ the equivalence classes of the elements of $\widetilde{\mathbb{Z}}$ with respect to the equivalence relation $$ z_{1} \equiv z_{2} \bmod I \Longleftrightarrow z_{1}-z_{2} \in I $$ The respective equivalence classes $\widetilde{\mathbb{Z}} / I$ are called residue classes modulo $I$. Problem 57. Check that residue classes define equivalence classes and that if $\widetilde{\mathbb{Z}}=\mathbb{Z}$ the notion of residue class coincides with congruence classes in modular ariphmetic. Problem 58. What is the number of elements of $\widetilde{\mathbb{Z}} / I$ for quadratic extensions? Problem 59. Prove that every $I \neq\{0\}$ contains a nonzero integer. Problem 60. Prove that the number of elements of $\widetilde{\mathbb{Z}} / I$ is finite. Problem 61. Pick $\alpha \in \widetilde{\mathbb{Z}} \backslash\{0\}$. Prove that there are only finitely many ideals $I$ such that $\alpha \in I$. Fix an integer basis $\left\{\alpha_{1}, \ldots, \alpha_{n}\right.$ of $\left.(1)=\widetilde{\mathbb{Z}}\right\}$. We attach to every $\alpha \in \widetilde{\mathbb{Z}}$ the sequence of integers $\left(x_{1}, \ldots, x_{n}\right)$ such that $\alpha=x_{1} \alpha_{1}+\ldots+x_{n} \alpha_{n}$. Definition. We say that such a sequence $\left(x_{1}, \ldots, x_{n}\right)$ is a vector. Set $\|\alpha\|=\left|x_{1}\right|+\left|x_{2}\right|+\ldots+\left|x_{n}\right|$. Problem 62. Prove that there exists $M_{1}>0$ such that, for every $\beta \in \widetilde{\mathbb{Z}} \backslash\{0\}$ there exists an integer basis $\left\{\beta_{1}, \ldots \beta_{n}\right.$ of the ideal $\left.(\beta)\right\}$ such that $\left\|\beta_{i}\right\|k$ such that $I^{k} \sim I^{m}$. b) Prove that there exists $\alpha \in \widetilde{\mathbb{Z}}$ such that $I^{m-k}=(\alpha)$. c) Show that for every nonzero ideal $I \subseteq \widetilde{\mathbb{Z}}$ there exist a nonzero ideal $J \subseteq \widetilde{\mathbb{Z}}$ and nonzero $\alpha \in \widetilde{\mathbb{Z}}$ such that $I J=(\alpha)$. Problem 70. Let $I, J \subset \widetilde{\mathbb{Z}}$ be nonzero ideals. Show that $J$ divides $I$ if and only if $I$ is contained in $J$. Problem 71. Show that every nonzero ideal $I \subseteq \widetilde{\mathbb{Z}}$ is a product of finitely many prime ideals in $\widetilde{\mathbb{Z}}$. Problem 72. Let $I, J, H$ be nonzero ideals in $\widetilde{\mathbb{Z}}$ such that $I$ and $J$ are relatively prime. Show that if $J H \subseteq I$ then $H \subseteq I$. Problem 73. Let $p_{1}, p_{2} \subset \widetilde{\mathbb{Z}}$ be prime ideals and $p_{1} \neq p_{2}$. Show that $p_{1}$ and $p_{2}$ are relatively prime. Problem 74. Let $I, J \subset \widetilde{\mathbb{Z}}$ such that $I$ is prime. Show that if $I^{m} \subseteq J$ then $J=I^{k}$ for some $k \leq m$. Problem 75. Show that any two powers of two relatively prime ideals are relatively prime. Problem 76. Prove that the prime ideal factorisation of $I$ in Problem 71 is unique up to a permutation of the factors. ## Fundamental Theorem of Arithmetic and Fermat's Last Theorem Fix $p>2$ and denote by $\zeta_{p}$ a complex $p$ th root of unity with $\zeta_{p} \neq 1$. The goal of this part of the project is to prove the following theorem. Theorem 8. Assume nonzero integers $x, y, z$ satisfy $x^{p}+y^{p}=z^{p}$ and the number of ideal classes of $\mathbb{Z}\left[\zeta_{p}\right]$ is not divisible by $p$. Then $p$ divides $x y z$. We hope that the participants of the project can prove Theorem 8 if they solve the following problems. Exercise 25. Prove that $1+\zeta_{p}+\zeta_{p}^{2}+\cdots+\zeta_{p}^{p-1}=0$. Problem 77. a) Provide a monic polynomial $f$ of degree $p-1$ with integer coefficients such that $1-\zeta_{p}$ is a root of $f$. b) Show that $p$ divides all the coefficients of this polynomial (except the first one). c) Prove that $f$ is irreducible in $\mathbb{Q}[x]$. Exercise 26. Consider rational numbers $a_{0}, \ldots, a_{p-1}, b_{0}, \ldots, b_{p-1}$. Show that $$ \sum_{i=0}^{p-1} a_{i} \zeta_{p}^{i}=\sum_{i=0}^{p-1} b_{i} \zeta_{p}^{i} $$ if and only if $$ a_{0}-b_{0}=a_{1}-b_{1}=\cdots=a_{p-1}-b_{p-1} $$ We say that $\gamma \in \overline{\mathbb{Z}}$ is invertible if $\gamma, 1 / \gamma \in \overline{\mathbb{Z}}$. Problem 78. Prove that $\frac{1-\zeta_{p}^{n}}{1-\zeta_{p}}$ is invertible if $p \nmid n$. Problem 79. Prove the following equalities of ideals $\left(1-\zeta_{p}\right)^{p}=(p)$. Exercise 27. Consider $\alpha \in \mathbb{Z}\left[\zeta_{p}\right]$. Show that if $p \mid \alpha$ then $\left(1-\zeta_{p}\right) \mid \alpha$. Let $\alpha_{0}, \ldots, \alpha=\alpha_{0}+\cdots+\alpha_{p-1} \zeta_{p}^{p-1}$. Problem 80. Prove that all coefficients of the following polynomial are rational: $$ P\left(a_{0}, \ldots, a_{p-1} ; x\right):=\Pi_{k=1}^{p-1}\left(x-\Sigma_{i=0}^{p-1} a_{i} \zeta_{p}^{k i}\right) $$ Recall that $P_{\alpha}(x)$ denotes the monic polynomial of minimal degree with coefficients in $\mathbb{R}$ such that $P_{\alpha}(\alpha)=0$. Problem 81* Prove that $P\left(a_{0}, \ldots, a_{p-1} ; x\right)=P_{a}(x)^{d}$ for some positive integer $d$. Problem 82. Assume $P_{\alpha}(x)$ has integer coefficients. Then for $1 \leq k \leq p-1$ and $l \in \mathbb{Z}$, the sum $\sum_{i=0}^{p-1} \alpha_{i} \zeta_{p}^{k i+l}$ is an algebraic integer. Problem 83. Assume $P_{\alpha}(x)$ has integer coefficients. Prove that $p\left(a_{i}-a_{j}\right) \in \mathbb{Z}$ for all $0 \leq i, j \leq p-1$. Problem 84. Show that $1 /\left(1-\zeta_{p}\right)$ is not an algebraic integer. Problem 85. Prove that $\mathbb{Z}\left[\zeta_{p}\right]$ coincides with the set of algebraic integers of $\mathbb{Q}\left[\zeta_{p}\right]$. Hint: use Exercise 27. Problem 86. Show that there exists $b \in \mathbb{Z}$ such that $p \mid\left(\alpha^{p}-b\right)$. Problem 87. a) Prove that $\sqrt{-1} \notin \mathbb{Z}\left[\zeta_{p}\right]$. b) Consider an odd prime $q \neq p$. Denote by $\zeta_{q}$ a $q$ th root of unity with $q \neq 1$. Show that $\zeta_{q} \notin \mathbb{Z}\left[\zeta_{p}\right]$. c) Denote by $\zeta_{p^{2}}$ a root of unity of degree $p^{2}$ with $\zeta_{p^{2}}^{p} \neq 1$. Show that $\zeta_{p^{2}} \notin \mathbb{Z}\left[\zeta_{p}\right]$. d) Find all roots of unity in $\mathbb{Z}\left[\zeta_{p}\right]$. Problem 88. Prove that for every invertible element $u \in \mathbb{Z}\left[\zeta_{p}\right]$, there exist $i \in \mathbb{Z}$ and $v \in \mathbb{R}$ such that $u=\zeta_{p}^{i} v$. Let $x, y, z, p$ be as in Theorem 8 . Problem 89. Show that ideals $\left(x+\zeta_{p}^{i} y\right)$, for $0 \leq i \leq p-1$, are relatively prime to each other. Problem 90. Prove Theorem 8. ## References [IR] K. Ireland, M. Rosen, A classical Introduction to Modern Number Theory, Graduate Texts in Mathematics 84, Springer-Verlag. [Mi] J. Milne, Algebraic number theory, http://jmilne.org/math/CourseNotes/ANT.pdf. [C1] K. Conrad, Factoring in quadratic fields, http://www.math.uconn.edu/ kconrad/blurbs/gradnumthy/quadraticgrad.pdf. [C2] K. Conrad, Ideal factorization, http://www.math.uconn.edu/ kconrad/blurbs/gradnumthy/idealfactor.pdf. [Wa] L. Washington, Introduction to Cyclotomic Fields, Graduate Texts in Mathematics 83, Springer-Verlag, 1996. [Ро] М. Постников, Теорема Ферма, Наука, 1978,. [Go] А. Гончаров, Арифметика гауссовых чисел, Журнал "Квант" 12 (1985), http://kvant.mccme.ru/1985/12/. [SS] В. Сендеров, А. Спивак, Сумми квадратов и целье гауссовы числа, Журнал "Квант" 3 (1993), см. также http://kvant.mccme.ru/pdf/1999/03/. [ZSS] А. Заславский, А. Скопенков, М. Скопенков (редакторы), Элементы математики в задачах - через олимпиады и кружки к профессии, 2-ое изд., издательство МЦНМО, 2017. ## Замечательные точки многоугольников А.Заславский, О.Заславский, П.Кожевников, Б.Френкин ${ }^{1}$ Был замысел хорош на диво, Но рока не перехитришь. Распорядился он глумливо, И от горы родилась мышъ. А.Великий Хорошо известно, что каждый треугольник обладает множеством более или менее интересных свойств. Например, количество так называемых замечательных точек в энциклопедии [1] уже перевалило за 10000. Эти точки определяют ряд прямых окружностей и других связанных с треугольником объектов. Цель проекта (мечта идиота) - найти аналоги этих объектов в произвольном многоугольнике. Разумеется, реализована она будет в очень малой степени, зато у участников Конференции останется значительный простор для собственных исследований. ## 1 Центры тяжести многоугольников Центром тяжести треугольника принято называть точку $M$ пересечения его медиан. Действительно, если расположить в вершинах треугольника равные массы, то их центр тяжести $M_{0}$ совпадет с $M$. В этой же точке находится центр тяжести $M_{2}$ треугольника, вырезанного, например, из картона. Однако, для треугольника, сделанного из проволоки, центром тяжести будет другая точка, обозначим ее $M_{1}$. Найти ее можно, воспользовавшись следующим общим свойством центров тяжести. Основное свойство. Пусть фигура $F$ является объединением двух непересекающихся фигур $F^{\prime}$ и $F^{\prime \prime}$. Тогда центр тяжести $M$ фигуры $F$ лежит на отрезке $M^{\prime} M^{\prime \prime}$, где $M^{\prime}, M^{\prime \prime}$ - центры тяжести $F^{\prime}, F^{\prime \prime}$, причем отношение отрезков $M M^{\prime} / M M^{\prime \prime}$ равно отношению $m_{2} / m_{1}$ масс $F^{\prime \prime}$ и $F^{\prime}$. При этом, если в качестве фигур $F^{\prime}, F^{\prime \prime}$ рассматриваются ломаные, то масса фигуры равна ее длине, а для плоских фигур массы равны их площадям. Строгие определения центров тяжести приведены в приложении. 1.1. Найдите центр тяжести $M_{1}$ проволочного треугольника. 1.2. Докажите, что точка $M_{1}$, точка пересечения медиан $M_{0}$ и центр $I$ вписанной окружности $A B C$ лежат на одной прямой (прямая Нагеля), причем $M_{0}$ делит отрезок $I M_{1}$ в отношении $2: 1$. 1.3. Докажите, что $M_{1}$ - радикальный центр трех вневписанных окружностей $A B C$, т.е. касательные, проведенные из $M_{1}$ к этим окружностям равны.[^2] 1.4. Докажите, что каждая из прямых $A_{0} M_{1}, B_{0} M_{1}, C_{0} M_{1}$, где $A_{0}, B_{0}, C_{0}-$ середины сторон $B C, C A, A B$ соответственно, делит периметр треугольника $A B C$ пополам. Итак для треугольника можно определить два центра тяжести $M_{0}$ и $M_{1}$, между которыми существует определенная связь. У произвольного многоугольника может существовать уже три центра тяжести: центр тяжести $M_{0}$ его вершин, центр тяжести $M_{1}$ периметра и центр тяжести $M_{2}$ сплошного многоугольника (для треугольника $M_{2}=M_{0}$ ). 1.5. Для четырехугольника $A B C D$ найдите точки $M_{0}$ и $M_{2}$. 1.6. Докажите, что $M_{0}$ лежит на отрезке $L M_{2}$, где $L-$ точка пересечения диагоналей четырехугольника, и делит его в отношении $3: 1$. 1.7. Для четырехугольника $A B C D$ найдите точку $M_{1}$. Похоже, что в общем случае точка $M_{1}$ не обладает никакими интересными свойствами. Но, если в четырехугольник $A B C D$ можно вписать окружность, то ситуация меняется. 1.8 . a) Докажите, что $M_{2}$ лежит на отрезке $I M_{1}$, где $I$ центр вписанной окружности, и делит его в отношении $2: 1$. b) Докажите, что это свойство верно для любого описанного многоугольника. 1.9. Докажите, что в любом четырехугольнике точка $M_{0}$ является серединой отрезка $M_{1} W$, где $W$ - середина $I L$. Будем теперь рассматривать четырехугольник, который является не только описанным, но и вписанным. По теореме Понселе можно, зафиксировав описанную и вписанную окружности, "вращать" четырехугольник между ними. 1.10. Какие кривые описывают при этом центры тяжести? ## Приложение. Определение центров тяжести Сдавать решения приведенных здесь упражнений необязательно, но за каждое сданное решение участник получает бонус - право рассказать решение одной из задач проекта устно. Определение 0. Материальной точкой называется пара $(X, m)$, где $X$ точка плоскости, а $m$ - положительное число ("масса" точки). Определение 1. Центром масс материальных точек $\left(X_{1}, m_{1}\right), \ldots,\left(X_{n}, m_{n}\right)$ называется такая точка $M$, что $$ m_{1} M \vec{X}_{1}+\cdots+m_{n} M \vec{X}_{n}=\overrightarrow{0} $$ Упражнение 1. Докажите существование и единственность центра масс. Упражнение 2. Докажите, что для любой точки $O$ $$ \operatorname{vec} O M=\frac{m_{1} O \vec{X}_{1}+\cdots+m_{n} O \vec{X}_{n}}{m_{1}+\cdots+m_{n}} $$ Упражнение 3. Докажите, что центр масс материальных точек $\left(A, m_{1}\right)$ и $\left(B, m_{2}\right.$ ) лежит на отрезке $A B$ и делит его в отношении $m_{2}: m_{1}$. Упражнение 4. Докажите, что центром масс материальных точек $(A, 1)$, $(B, 1),(C, 1)$ является точка пересечения медиан треугольника $A B C$. Упражнение 5. Докажите, что центр масс материальных точек $\left(X_{1}, m_{1}\right), \ldots$, $\left(X_{n}, m_{n}\right),\left(X_{n+1}, m_{n+1}\right)$ совпадает с центром масс материальных точек $\left(M, m_{1}+\right.$ $\left.\cdots+m_{n}\right),\left(X_{n+1}, m_{n+1}\right)$, где $M$ - центр масс материальных точек $\left(X_{1}, m_{1}\right), \ldots$, $\left(X_{n}, m_{n}\right)$. Определение 2. Центром масс $n$ отрезков, каждые два из которых имеют не более одной общей точки называется центр масс материальных точек $\left(M_{1}, l_{1}\right), \ldots,\left(M_{n}, l_{n}\right)$, где $M_{i}$ - середина $i$-го отрезка, а $l_{i}$ - его длина. Определение 3. Пусть фигура $F$ является объединением $n$ треугольников, никакие два из которых не имеют общих внутренних точек. Центром масс фигуры $F$ называется центр масс материальных точек $\left(M_{1}, S_{1}\right), \ldots,\left(M_{n}, S_{n}\right)$, где $M_{i}$ - точка пересечения медиан $i$-го треугольника, а $S_{i}$ - его площадь. Упражнение 6. Докажите, что при любом разбиении многоугольника на треугольники центром масс объединения этих треугольников будет одна и та же точка (эта точка называется центром масс многоугольника). Упражнение $7^{*}$. Пусть прямые $A B$ и $C D$ пересекаются в точке $E$, а прямые $A D$ и $B C$ - в точке $F$. Докажите, что середины отрезков $A C, B D$ и $E F$ лежат на одной прямой (прямая Гаусса четырехугольника $A B C D$ ). В проекте под центром тяжести $M_{0}$ многоугольника $A_{1} \ldots A_{n}$ подразумевается центр масс материальных точек $\left(A_{1}, 1\right), \ldots,\left(A_{n}, 1\right)$, под центром тяжести $M_{1}$ - центр масс отрезков $A_{1} A_{2}, \ldots, A_{n-1} A_{n}, A_{n} A_{1}$, под центром тяжести $M_{2}-$ центр масс многоугольника. ## 2 Прямые Эйлера и Нагеля Известно, что в любом треугольнике центр описанной окружности $O$, центр тяжести $M_{0}$ и ортоцентр $H$ лежат на одной прямой, которая называется прямой Эйлера, причем $M_{0}$ делит отрезок $O H$ в отношении $1: 2$. Также, если $A_{1}$, $B_{1}, C_{1}$ - точки касания вписанной окружности со сторонами $B C, C A, A B$, а точки $A_{2}, B_{2}, C_{2}$ симметричны $A_{1}, B_{1}, C_{1}$ относительно середин соответствующих сторон (в этих точках стороны касаются соответствующих вневписанных окружностей), то прямые $A A_{2}, B B_{2}, C C_{2}$ пересекаются в одной точке $N$, которая называется точкой Нагеля. При этом $M_{0}$ лежит на отрезке $I N$ и делит его в отношении $1: 2$. Отметим также, что каждая из прямых $A A_{2}, B B_{2}, C C_{2}$ делит периметр треугольника пополам. Наша цель - найти аналоги прямой Эйлера для вписанного многоугольника и прямой Нагеля для описанного. 2.1. (А.Мякишев, II Олимпиада им. И.Ф.Шарыгина) Пусть четырехугольник $A B C D$ вписан в окружность с центром $O ; H_{a}, H_{b}, H_{c}, H_{d}$ - ортоцентры треугольников $B C D, C D A, D A B, A B C$ соответственно; $H$ - точка пересечения прямых $H_{a} H_{c}$ и $H_{b} H_{d}$. Докажите, что центр тяжести $M_{2}$ лежит на отрезке $O H$ и делит его в отношении $1: 2$. 2.2. (А.Мякишев, II Олимпиада им. И.Ф.Шарыгина) Пусть четырехугольник $A B C D$ описан около окружности с центром $I$; точки $T, U, V, W$ симметричны точкам касания окружности со сторонами $A B, B C, C D, D A$ относительно середин этих сторон. a) Докажите, что каждая из прямых $T V$ и $U W$ делит периметр четырехугольника пополам. b) Пусть $N$ - точка пересечения прямых $T V$ и $U W$. Докажите, что $M_{2}$ лежит на отрезке $I N$ и делит его в отношении $1: 2$. Другой подход к определению прямой Эйлера предложил И.Романов [4]. Определим ортоцентр вписанного $n$-угольника $A_{1} \ldots A_{n}$ по индукции. Пусть $H_{1}, \ldots, H_{n}$ - ортоцентры ( $n-1$ )-угольников $A_{2} \ldots A_{n}, \ldots, A_{1} \ldots A_{n-1}$ соответственно. 2.3. Докажите, что прямые $A_{1} H_{1}, \ldots, A_{n} H_{n}$ пересекаются в одной точке. 2.4. Назовем полученную в предыдущей задаче точку $H$ ортоцентром $n$ угольника. Докажите, что центр тяжести $M_{0}$ лежит на отрезке $O H$ и делит его в отношении $(n-2): 2$. 2.5. Пусть $A B C D$ - произвольный четырехугольник. Рассматриваем два обобщения ортоцентра: $H^{*}$ - центр параллелограмма, образованного ортоцентрами треугольников $A B L, B C L, C D L, D A L$. $H^{* *}=H_{a} H_{c} \cap H_{b} H_{d}$, где, как обычно, $H_{a}$ - ортоцентр треугольника $B C D$, и т.д.. Далее обобщаем $O$ как $O^{* *}=O_{a} O_{c} \cap O_{b} O_{d}$, где $O_{a}$ - центр описанной окружности треугольника $B C D$, и т.д.. (иначе говоря, $O^{* *}$ - пересечение серединных перпендикуляров к $A C$ и $B D$ ). Докажите, что а) $M_{0}$ - середина $O^{* *} H^{*}$; b) (Я. Ганин, А. Мякишев) $M_{2}$ лежит на отрезке $O^{* *} H^{* *}$ и делит его в отношении $1: 2$. c) $H^{*}$ - середина $L H^{* *}$. ## 3 Квазицентры описанной и вписанной окружностей В этой части мы попытаемся для произвольного четырехугольника определить точки $O, I$, обладающие свойствами центров описанной и вписанной окружностей. Разумеется, для вписанного (описанного) четырехугольника точка $O(I)$ должна совпадать с центром описанной (вписанной) окружности. 3.1. Докажите, что для любого вписанно-описанного четырехугольника $A B C D$ центры $O, I$ и точка пересечения диагоналей $L$ лежат на одной прямой. 3.2. Пусть диагонали четырехугольника $P Q R S$ пересекаются в точке $I$; точки $A, B, C, D$ - проекции $I$ на $P Q, Q R, R S, S P$ соответственно. Докажите, что a) четырехугольник $A B C D$ описанный тогда и только тогда, когда четырехугольник $P Q R S$ вписанный; b) если четырехугольник $A B C D$ описанный, то $I$ - центр его вписанной окружности; Пусть прямые $I A, I B, I C, I D$ пересекают $R S, S P, P Q, Q R$ соответственно в точках $A^{\prime}, B^{\prime}, C^{\prime}, D^{\prime}$. 3.3. Докажите, что a) четырехугольник $A B C D$ вписанный тогда и только тогда, когда $P R \perp$ $Q S$; b) если $P R \perp Q S$, то $A^{\prime} B^{\prime} C^{\prime} D^{\prime}$ - прямоугольник и точки $A, B, C, D, A^{\prime}, B^{\prime}$, $C^{\prime}, D^{\prime}$ лежат на одной окружности. 3.4. Восстановите четырехугольник $P Q R S$ по точкам $A, B, C, D$, если известно, что точка $I$ лежит внутри четырехугольника $A B C D$. Определение. Назовем квазицентром вписанной окружности выпуклого четырехугольника $A B C D$ построенную в предыдущей задаче точку $I$, а квазицентром описанной окружности точку $O$ пересечения прямых $A^{\prime} C^{\prime}$ и $B^{\prime} D^{\prime}$. (Будем считать, что $I$ лежит внутри $A B C D$ ) 3.5. Докажите, что квазицентры $O, I$ и точка пересечения диагоналей $L$ лежат на одной прямой. 3.6. Для вписанно-описанного четырехугольника выразите радиусы описанной и вписанной окружностей через длины отрезков $O I$ и $O L$. Последняя задача позволяет определить для произвольного четырехугольника квазивписанную и квазиописанную окружности. Пока неизвестно, обладают ли эти окружности какими-либо интересными свойствами. Опишем другой подход к определению квазицентров. 3.7. Пусть $I_{a}, I_{b}, I_{c}$ - центры вневписанных окружностей треугольника $A B C ; J$ - центр описанной окружности треугольника $I_{a} I_{b} I_{c}$. Докажите, что $O$ - середина отрезка $I J$. 3.8. Докажите, что для произвольного четырехугольника биссектрисы его внутренних углов образуют вписанный четырехугольник и биссектрисы внешних углов также образуют вписанный четырехугольник. Обозначим центры окружностей четырехугольников, образованных биссектрисам внутренних и внешних углов через $I$ и $J$ соответственно. 3.9. (VII Олимпиада им. И.Ф.Шарыгина) Докажите, что для четырехугольника, вписанного в окружность с центром $O$ точки $I$ и $J$ симметричны относительно $O$. Теперь в качестве квазицентров вписанной и описанной окружностей можно взять точку $I$ и середину $O$ отрезка $I J$ соответственно. $\mathrm{K}$ сожалению, определенная таким образом прямая $O I$ может не проходить через точку пересечения диагоналей $L$. Еще один подход к определению квазицентра описанной окружности предложен в [6]. 3.10. Пусть прямые $A B$ и $C D$ пересекаются в точке $X, A D$ и $B C$ - в точке $Y, A C$ и $B D$ - в точке $Z ; M_{X}$ - точка Микеля прямых $A D, B C, A C$ и $B D$, $M_{Y}$ - точка Микеля прямых $A B, B D, A C$ и $B D, M_{Z}$ - точка Микеля прямых $A D, B C, A B$ и $C D$. Докажите, что a) прямые $X M_{X}, Y M_{Y}$ и $Z M_{Z}$ пересекаются в одной точке; b) если точки $A, B, C, D$ лежат на одной окружности, то эти прямые пересекаются в ее центре. Соответственно полученную в последней задаче точку также можно считать квазицентром описанной окружности. ## 4 Дополнительные задачи 4.1. Пусть $A B C D$ - четырехугольник, не имеющий параллельных сторон, описанный около окружности с центром $I$. Точки $X, Y, Z, T$ - точки касания окружности со сторонами $A B, B C, C D, D A$ соответственно. Как всегда, $L=$ $A C \cap B D$ (а также $L=X Z \cap Y T$ ). $X^{\prime}$ симметрична $X$ относительно середины $M_{A B}$ стороны $A B ; Y^{\prime}, Z^{\prime}, T^{\prime}$ определяются аналогично. $N=X^{\prime} Z^{\prime} \cap Y^{\prime} T^{\prime}$ - точка Нагеля. Докажите, что условие $M_{0}=I$ эквивалентно следующим условиям: a) $A X+C Z=B Y+D T$ b) $X Z \| X^{\prime} Z^{\prime}$ (или $X Z \| M_{A B} M_{C D}$ ); c) $X^{\prime}, Z^{\prime}$ и $B C \cap A D$ лежат на одной прямой; d) $L, I, N$ лежат на одной прямой; е) (А. Заславский, М. Исаев, Д. Цветов, Всероссийская олимпиада 2005 г.) $I A \cdot I C=I B \cdot I D$. 4.2. (А.Мякишев) Треугольники $A B C$ и $A^{\prime} B^{\prime} C^{\prime}$ называются ортологичными, если перпендикуляры, опущенные из $A^{\prime}, B^{\prime}, C^{\prime}$ соответственно на $B C, C A$, $A B$, пересекаются в одной точке. Четырехугольники $A B C D$ и $A^{\prime} B^{\prime} C^{\prime} D^{\prime}$ называются ортологичными, если ортологичны треугольники $A B C$ и $A^{\prime} B^{\prime} C^{\prime}, B C D$ и $B^{\prime} C^{\prime} D^{\prime}, C D A$ и $C^{\prime} D^{\prime} A^{\prime}, D A B$ и $D^{\prime} A^{\prime} B^{\prime}$. Пусть четырехугольники $A B C D$ и $A^{\prime} B^{\prime} C^{\prime} D^{\prime}$ ортологичны, диагонали $A C$ и $B D$ пересекаются в точке $L, A^{\prime} C^{\prime}$ и $B^{\prime} D^{\prime}$ - в точке $L^{\prime}$. Докажите, что $A L: L C=A^{\prime} L^{\prime}: L^{\prime} C^{\prime}$ и $B L: L D=B^{\prime} L^{\prime}: L^{\prime} D^{\prime}$ (т.е. ортологичные четырехугольники аффинно эквивалентны). ## Список литературы [1] http://faculty.evansville.edu/ck6/encyclopedia/ETC.html. [2] Ф.Ивлев. Центры тяжести многоугольников. Доклад на ММКШ. 2008. https://www.mccme.ru/circles/oim/mmks/notes.htm [3] A.Akopyan. Some remarks on the circumcenter of mass. https://arxiv.org/pdf/1512.08655.pdf [4] И.Романов. Прямая Эйлера n-угольника. Доклад на ММКШ. 2017. https://www.mccme.ru/circles/oim/mmks/works2017/ignatov2.pdf [5] А.Заславский. Диагонально-перпендикулярное отображение четырехугольников. Квант. 1998. №4. [6] M.Rolnek, Le Anh Dung. The Miquel Points, Pseudocircumcenter, and EulerPoncelet Point of a Complete Quadrilateral. Forum Geometricorum V. 14 (2014). https://personal.us.es/rbarroso/trianguloscabri/sol/FG201413.pdf. ## Замечательные точки многоугольников Решения ## 1 Центры тяжести многоугольников 1.1. Ответ. Центр окружности, вписанной в треугольник $A_{0} B_{0} C_{0}$. Доказательство. Так как точки $A_{0}, B_{0}, C_{0}$ являются серединами отрезков $B C, C A, A B$, поместим в них массы, равные длинам этих отрезков. Тогда центром тяжести масс в точках $A_{0}, B_{0}$ будет точка, делящая отрезок $A_{0} B_{0}$ в отношении $A C: B C=A_{0} C_{0}: B_{0} C_{0}$, т.е. основание биссектрисы треугольника $A_{0} B_{0} C_{0}$. Следовательно, центр тяжести всех трех масс лежит на этой биссектрисе. Аналогично получаем, что он лежит и на других биссектрисах треугольника $A_{0} B_{0} C_{0}$ и, значит, совпадает с центром вписанной в этот треугольник окружности (рис.1). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-062.jpg?height=653&width=886&top_left_y=1324&top_left_x=565) Рис. 1 1.2. Следует из предыдущей задачи и гомотетичности треугольников $A B C$, $A_{0} B_{0} C_{0}$ относительно $I$. 1.3. Пусть вневписанные окружности, касающиеся сторон $A C$ и $B C$, касаются продолжений стороны $A B$ в точках $X, Y$. Тогда $B X=A Y=p$ (полупериметр треугольника) и, значит, $C_{0} X=C_{0} Y$. Кроме того, линия центров этих окружностей перпендикулярна биссектрисе угла $C$, а значит, и параллельной ей прямой $C_{0} M_{1}$. Таким образом, $M_{1}$ лежит на радикальной оси этих окружностей. Аналогично получаем, что $M_{1}$ лежит на радикальной оси любой другой пары вневписанных окружностей. 1.4. Пусть, например, $C_{0} M_{1}$ пересекает сторону $A C$ в точке $X$. Из предыдущей задачи следует, что $X$ делит пополам отрезок между точками касания прямой $A C$ с вневписанными окружностями, касающимися сторон $A C$ и $B C$ (рис.2). Поскольку расстояния от этих точек до вершины $A$ равны соответственно $p-c$ и $p$, то $A X=p-c / 2$ и $A X+A C_{0}=p$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-063.jpg?height=1065&width=1313&top_left_y=995&top_left_x=377) Рис. 2 1.5. Ответ. $M_{0}$ - центр параллелограмма $P Q R S$, где $P, Q, R, S$ - середины сторон $A B, B C, C D, D A . M_{2}$ - точка пересечения прямой $l_{1}$, соединяющей центры тяжести треугольников $A B C$ и $A D C$, с прямой $l_{2}$, соединяющей центры тяжести треугольников $A B D$ и $B C D$. 1.6. Пусть $U, V$ - середины диагоналей $A C$ и $B D, L$ - точка их пересечения. Центр тяжести треугольника $A B C$ лежит на его медиане $B U$ и делит ее в отношении $2: 1$. Аналогично центр тяжести треугольника $A C D$ лежит на $D U$ и делит ее в том же отношении. Прямая $l_{1}$, проходящая через эти центры, параллельна диагонали $B D$ и пересекает диагональ $A C$ в точке, которая делит отрезок $U L$ в отношении $1: 2$. Аналогично $M_{2}$ лежит на прямой, параллельной $A C$ и делящей отрезок $V L$ в отношении $1: 2$. Отметим, что если провести прямые, параллельные $A C$ и $B D$, через $M_{0}$, то они разделят отрезки $U L$ и $V L$ пополам. Отсюда вытекает утверждение задачи (рис.3). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-064.jpg?height=794&width=940&top_left_y=991&top_left_x=498) Рис. 3 1.7. Так как $P$ и $Q$ - центры тяжести отрезков $A B$ и $B C$, центр тяжести $X_{1}$ объединения этих отрезков лежит на $P Q$ и $P X_{1} / Q X_{1}=B C / A B$. Построить эту точку можно следующим образом: проведем биссектрису $B B^{\prime}$ треугольника $B P Q$ и найдем точку, симметричную $B^{\prime}$ относительно середины отрезка $P Q$. Аналогично строятся центры тяжести $X_{2}, Y_{1}, Y_{2}$ ломаных $C D A, D A B, B C D$. $M_{1}$ - это точка пересечения прямых $X_{1} X_{2}$ и $Y_{1} Y_{2}$ (рис.4). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-065.jpg?height=760&width=965&top_left_y=471&top_left_x=474) Рис. 4 1.8 . a) Центры тяжести (сплошных) треугольников $I A B, I B C, I C D, I D A$, где $I$ - центр вписанной окружности, делят их медианы $I P, I Q, I R, I S$ в отношении $2: 1$, т.е. образованный ими четырехугольник гомотетичен $P Q R S$ с центром $I$ и коэффициентом $\frac{2}{3}$. Так как площади треугольников $I A B, I B C$, $I C D, I D A$ относятся так же, как соответствующие стороны $A B C D, M_{1}$ при этой гомотетии переходит в $M_{2}$. b) Доказательство аналогично. 1.9. Следует из двух предыдущих задач и теоремы о центрах трех гомотетий. 1.10. Указание. Точка $M_{0}$ - середина отрезка между серединами диагоналей четырехугольника. Используя тот факт, что прямая Гаусса описанного четырехугольника проходит через центр вписанной окружности, нетрудно вывести, что траектория $M_{0}$ - окружность. Теперь, применив соответствующие гомотетии, получаем, что и траектории двух других центров - окружности (рис.5). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-066.jpg?height=994&width=991&top_left_y=377&top_left_x=567) Рис. 5 ## 2 Прямые Эйлера и Нагеля 2.1. Пусть $M_{a}$ и $H_{a}$ - соответственно центроид (точка $M_{2}$ ) и ортоцентр треугольника $B C D$. Центроиды и ортоцентры остальных трех треугольников обозначим аналогично. Все треугольники имеют общую описанную окружность с центром в $O$. Рассмотрев прямые Эйлера этих треугольников, заметим, что четырехугольник $M_{a} M_{b} M_{c} M_{d}$ переходит в четырехугольник $H_{a} H_{b} H_{c} H_{d}$ при гомотетии с центром в $O$ и коэффициентом 3. Соответственно, точки пересечения диагоналей этих четырехугольников переходят друг в друга. 2.2 . a) Из определения следует, что $A T+D V=B C, B T+C V=A D$, т.е. $T A+$ $A D+D V=V C+C B+B T$. b) Пусть $a, b, c, d$ - длины касательных к вписанной окружности из вершин $A, B, C, D$. Очевидно, что, если поместить в $A, B, C, D$ массы $a, b, c, d$, то центром тяжести полученной системы будет точка $N$, а, если поместить в вершины массы $2 a+b+d, 2 b+a+c, 2 c+b+d, 2 d+c+a$, то - точка $M_{1}$. Осталось показать, что $I$ - центр тяжести масс $b+d, a+c, b+d, a+c$ и воспользоваться утверждением задачи 1.8 . Точка $I$ удовлетворяет соотношению $S_{I A B}-S_{I B C}+S_{I C D}-S_{I D A}=0$. Этому же соотношению удовлетворяют середины $U$ и $V$ диагоналей четырехугольника. Следовательно, эти три точки лежат на одной прямой (это утверждение называется теоремой Монжа). Пусть теперь $X, Y$ - точки касания вписанной окружности со сторонами $B C$ и $A D$. Тогда прямая $X Y$ образует равные углы с этими сторонами и по теореме Брианшона проходит через точку $L$ пересечения диагоналей. Применив теорему синусов к треугольникам $L X B$ и $L Y D$, получим, что $B L / D L=b / d$. Аналогично, $A L / C L=a / c$. Отсюда и из соотношений $S_{U B C} / S_{U A D}=B L / D L, S_{V B C} / S_{V A D}=C L / A L, S_{I B C} / S_{I A D}=(b+c) /(a+d)$ вытекает, что $I$ делит отрезок $A C$ в отношении $(a+c) /(b+d)$, что и требуется. 2.3. Указание. Примените индукцию. 2.4. Указание. Примените индукцию. 2.5. а) Пусть $M_{A C}, M_{B D}$ - середины диагоналей $A C, B D$. Очевидно, что $O^{* *} M_{A C} \perp A C$ и $O^{* *} M_{B D} \perp B D$. С другой стороны $H^{*} M_{A C} \perp B D$ и $H^{*} M_{B D} \perp$ $A C$, потому что $H^{*}$ - центр параллелограмма, стороны которого перпендикулярны диагоналям четырехугольника и проходят через его вершины. Следовательно, $H^{*} M_{A C} O^{* *} M_{B D}$ - параллелограмм, откуда и следует утверждение задачи. b) Указание. Воспользуйтесь утверждением задачи 4.2. c) Очевидно следует из двух предыдущих пунктов. ## 3 Квазицентры описанной и вписанной окружностей 3.1. Воспользуемся следующим утверждением. Лемма. Пусть точка $P$ лежит внутри окружности с центром $O$. Два перпендикулярных луча с началом $P$ пересекают окружность в точках $X, Y$. Тогда геометрическим местом точек пересечения касательных к окружности в $X$ и $Y$ будет окружность с центром, лежащим на $O P$. Доказательство. Пусть $Z$ - четвертая вершина прямоугольника $P X Z Y$. Так как $O P^{2}+O Z^{2}=O X^{2}+O Y^{2}, Z$ описывает окружность с центром $O$. Значит, середина отрезка $X Y$ описывает окружность, центром которой является середина $O P$, и инверсная к ней точка пересечения касательных также описывает окружность. Теперь утверждение задачи следует из того факта, что прямые, соединяющие точки касания противоположных сторон четырехугольника с вписанной окружностью, перпендикулярны и проходят через точку пересечения его диагоналей. 3.2. Так как четырехугольники $I A Q B, I B R C$ вписанные, то $\angle I B A=\angle I Q A$, $\angle I B C=\angle I R C$, откуда следуют оба утверждения. 3.3. Доказательство аналогично предыдущей задаче. 3.4. Будем считать, что лучи $P Q$ и $S R$ пересекаются в точке $X$. Тогда $\angle P X S=\angle P I S-\angle I P A-\angle C S I$. Из вписанности четырехугольников IAPD, $I C S D$ получаем, что $\angle I P A+\angle C S I=\angle C D A$. Кроме того, $\angle P I S=\angle R I Q=$ $(\angle P A D+\angle D C S+\angle R C B+\angle B A Q) / 2=(\angle A B C+\angle C D A) / 2$. Следовательно, $\angle A I C=\pi-\angle P X S=\pi-(\angle A B C-\angle C D A) / 2$ (рис.6). Найдя аналогично угол $B I D$ мы сможем построить точку $I$ как точку пересечения соответствующих дуг окружностей, а затем и четырехугольник $P Q R S$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-069.jpg?height=622&width=1462&top_left_y=497&top_left_x=434) Рис. 6 3.5. Указание. Примените центральную проекцию, переводящую точки $P$, $Q, R, S$ в вершины параллелограмма. 3.6. Ответ. $R^{2}=\frac{O L \cdot O I^{2}}{2 O I-O L}, r^{2}=\frac{\left(R^{2}-O I^{2}\right)^{2}}{2\left(R^{2}+O I^{2}\right)}$. Указание. По теореме Понселе достаточно рассмотреть четырехугольник, одна из диагоналей которого является диаметром описанной окружности. 3.7. Указание. Докажите, что отрезки $I_{a} J, I_{b} J, I_{c} J$ перпендикулярны соответствующим сторонам треугольника $A B C$. 3.8. Доказывается непосредственным вычислением углов. 3.9. Пусть биссектрисы углов $A$ и $B$ пересекаются в точке $K, B$ и $C$ - в точке $L, C$ и $D$ - в точке $M, D$ и $A$ - в точке $N$ (рис.9.4). Тогда прямая $K M-$ биссектриса угла между $A D$ и $B C$. Обозначив этот угол через $\phi$, по теореме о внешнем угле получаем, что $\angle L K M=\angle B / 2-\phi / 2=(\pi-\angle A) / 2=\angle C / 2$ и, значит, $\angle L I M=\angle C$. С другой стороны, перпендикуляры из $L$ на $B C$ и из $M$ на $C D$ образуют с $M L$ углы, равные $(\pi-\angle C) / 2$, т.е. треугольник, образованный этими перпендикулярами и $M L,-$ равнобедренный с углом при вершине, равным углу $C$. Поэтому вершина этого треугольника совпадает с $I$. Таким образом, перпендикуляры, опущенные из вершин четырехугольника $K L M N$ на соответствующие стороны $A B C D$, проходят через I (рис.7). Аналогично получаем, что перпендикуляры из вершин четырехугольника, образованного внеш- ними биссектрисами, проходят через $J$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-070.jpg?height=808&width=828&top_left_y=361&top_left_x=637) Рис. 7 Пусть теперь $K^{\prime}$ - точка пересечения биссектрис внешних углов $A$ и $B$. Так как четырехугольник $A K B K^{\prime}$ вписан в окружность с диаметром $K K^{\prime}$, то проекции $K$ и $K^{\prime}$ на $A B$ симметричны относительно середины $A B$. Отсюда и из утверждения, доказанного выше, следует, что проекции $I$ и $J$ на каждую из сторон $A B C D$ симметричны относительно середины этой стороны, что равносильно утверждению задачи. 3.10. См. [6]. ## Список литературы [1] http://faculty.evansville.edu/ck6/encyclopedia/ETC.html. [2] Ф.Ивлев. Центры тяжести многоугольников. Доклад на ММКШ. 2008. https://www.mccme.ru/circles/oim/mmks/notes.htm [3] A.Akopyan. Some remarks on the circumcenter of mass. https://arxiv.org/pdf/1512.08655.pdf [4] И.Романов. Прямая Эйлера n-угольника. Доклад на ММКШ. 2017. https://www.mccme.ru/circles/oim/mmks/works2017/ignatov2.pdf [5] А.Заславский. Диагонально-перпендикулярное отображение четырехугольников. Квант. 1998. №4. [6] M.Rolnek, Le Anh Dung. The Miquel Points, Pseudocircumcenter, and EulerPoncelet Point of a Complete Quadrilateral. Forum Geometricorum V. 14 (2014). https://personal.us.es/rbarroso/trianguloscabri/sol/FG201413.pdf. ## Remarkable points of polygons A.Zaslavsky, B.Frenkin, P.Kozhevnikov, O.Zaslavsky ${ }^{1}$ It is well-known that any triangle has many various, more or less remarkable points. For instance, the number of so-called triangle centers in the encyclopedia [1] has already exceeded 10000. These points define various lines, circles and other objects related to the triangle. The aim of this project (our pipe dream) is to find analogues of these objects in an arbitrary polygon. Of course, it will be realized in the least, but on the other hand the participants of the Conference will have a large scope for their own research. ## 1 Barycenters of polygons The centroid of a triangle is the point $M$ of concurrence of its medians. Indeed, if we place equal weights in the vertices of a triangle, their barycenter $M_{0}$ will coincide with $M$. The same point is the barycenter $M_{2}$ of a triangle cut out from cardboard for example. However the barycenter of a triangle made of wire differs from these points. Denote it by $M_{1}$. It can be determined using the following general property of barycenters. The main property. Suppose some figure $F$ is a disjoint union of figures $F^{\prime}$ and $F^{\prime \prime}$. Then the barycenter $M$ of $F$ lies on the segment $M^{\prime} M^{\prime \prime}$, where $M^{\prime}, M^{\prime \prime}$ are the barycenters of $F^{\prime}$ and $F^{\prime \prime}$ respectively. Moreover, the ratio $M M^{\prime} / M M^{\prime \prime}$ equals the ratio $m_{2} / m_{1}$ of masses of $F^{\prime \prime}$ and $F^{\prime}$. Here, if the figures $F^{\prime}$ and $F^{\prime \prime}$ are piecewise linear curves, then the weight of a figure is proportional to its length, and for plane figures the weights are proportional to their areas. 1.1. Find the barycenter $M_{1}$ of a triangle made of wire. 1.2. Prove that the point $M_{1}$, the point $M_{0}$ of concurrence of medians and the incenter $I$ of $A B C$ are collinear (the Nagel line), moreover $M_{0}$ divides $I M_{1}$ in the ratio $2: 1$. 1.3. Prove that $M_{1}$ is the radical center of three excircles of $A B C$, i.e., the segments of tangents from $M_{1}$ to these circles are equal. 1.4. Prove that each of the lines $A_{0} M_{1}, B_{0} M_{1}, C_{0} M_{1}$, where $A_{0}, B_{0}, C_{0}$ are the middle points of the segments $B C, C A, A B$ respectively, bisects the perimeter of $A B C$. So, for any triangle we can define two barycenters $M_{0}$ and $M_{1}$, and they are somehow connected to each other. An arbitrary polygon can have three barycenters: the barycenter $M_{0}$ of its vertices, the barycenter $M_{1}$ of the union of its sides and the barycenter $M_{2}$ of the whole polygon (in the degenerate case of a triangle we have $M_{2}=M_{0}$. 1.5. Determine points $M_{0}$ and $M_{2}$ for a quadrilateral $A B C D$. 1.6. Prove that $M_{0}$ lies on the segment $L M_{2}$, where $L$ is the intersection point of the diagonals of the quadrilateral, and $M_{0}$ divides $L M_{2}$ in the ratio $3: 1$.[^3] 1.7. Determine point $M_{1}$ for a quadrilateral $A B C D$. It seems that generally the point $M_{1}$ has no remarkable properties. However for a circumscribed $A B C D$ the situation is different. 1.8 . a) Show that $M_{2}$ lies on the segment $I M_{1}$, where $I$ is the incenter, and divides it in the ratio $2: 1$. b) Show that the same is true for any circumscribed polygon. 1.9. Show that in any quadrilateral, $M_{0}$ is the midpoint of the segment $M_{1} W$, where $W$ is the midpoint of $I L$. Now consider a quadrilateral which is not only circumscribed, but inscribed as well. Poncelet theorem then asserts that one can fix the incircle and the circumcircle and "rotate" the quadrilateral between them. 1.10. What is the locus of each barycenter of the quadrilateral? ## Appendix. Definition of barycenters It is not obligatory to deliver the solutions of exercises given here, but each delivered solution gives a bonus to the participant, namely the right to tell the solution of one problem of the project orally. Definition 0. Material point is a pair $(X, m)$, where $X$ is a point of the plane, and $m$ is a positive number ("the mass" of the point). Definition 1. The mass center of material points $\left(X_{1}, m_{1}\right), \ldots,\left(X_{n}, m_{n}\right)$ is the point $M$ such that $$ m_{1} \overrightarrow{M X_{1}}+\cdots+m_{n} \overrightarrow{M X_{n}}=\overrightarrow{0} $$ Exercise 1. Prove that there exists a unique mass center. Exercise 2. Prove that for any point $O$ $$ \overrightarrow{O M}=\frac{m_{1} \overrightarrow{O X_{1}}+\cdots+m_{n} \overrightarrow{O X_{n}}}{m_{1}+\cdots+m_{n}} $$ Exercise 3. Prove that the mass center of material points $\left(A, m_{1}\right)$ and $\left(B, m_{2}\right)$ lies on the segment $A B$ and dissects it in the ratio $m_{2}: m_{1}$. Exercise 4. Prove that the mass center of material points $(A, 1),(B, 1),(C, 1)$ coincides with the centroid of triangle $A B C$. Exercise 5. Prove that the mass center of material points $\left(X_{1}, m_{1}\right), \ldots,\left(X_{n}, m_{n}\right)$, $\left(X_{n+1}, m_{n+1}\right)$ coincides with the mass center of material points $\left(M, m_{1}+\cdots+m_{n}\right)$, $\left(X_{n+1}, m_{n+1}\right)$, where $M$ is the mass center of $\left(X_{1}, m_{1}\right), \ldots,\left(X_{n}, m_{n}\right)$. Definition 2. The mass center of $n$ segments having at most one common point pairwise is the mass center of material points $\left(M_{1}, l_{1}\right), \ldots,\left(M_{n}, l_{n}\right)$, where $M_{i}$ is the midpoint of $i$-th segment, and $l_{i}$ is its length. Definition 3. Let $F$ be the union of $n$ triangles having no common inner points pairwise. The mass center of $F$ is the mass center of material points $\left(M_{1}, S_{1}\right), \ldots,\left(M_{n}, S_{n}\right)$, where $M_{i}$ is the centroid of $i$-th triangle, and $S_{i}$ is its area. Exercise 6. Prove that for any division of a polygon into triangles, the mass center of the union of these triangles is the same point (called the mass center of the polygon). Exercise $7^{*}$. Let lines $A B$ and $C D$ meet at point $E$, and lines $A D$ and $B C$ meet at point $F$. Prove that the midpoints of segments $A C, B D$ and $E F$ are on a line (the Gauss line of quadrilateral $A B C D$ ). In this project the barycenter $M_{0}$ of a polygon $A_{1} \ldots A_{n}$ denotes the mass center of material points $\left(A_{1}, 1\right), \ldots,\left(A_{n}, 1\right)$, the barycenter $M_{1}$ denotes the mass center of segments $A_{1} A_{2}, \ldots, A_{n-1} A_{n}, A_{n} A_{1}$, and the barycenter $M_{2}$ denotes the mass center of the polygon. ## 2 Euler and Nagel lines It is well known that in any triangle the circumcenter $O$, the centroid $M_{0}$ and the orthocenter $H$ lie on a line, which is called Euler line. Moreover $M_{0}$ divides the segment $O H$ in the ratio $1: 2$. Furthermore let $A_{1}, B_{1}, C_{1}$ be the points of tangency of the incircle with the sides $B C, C A, A B$ respectively, and the points $A_{2}, B_{2}, C_{2}$ are symmetric to $A_{1}, B_{1}, C_{1}$ with respect to the midpoints of the corresponding sides (these are the points of tangency of the sides with the corresponding excircles). Then the lines $A A_{2}, B B_{2}, C C_{2}$ concur at the point $N$ which is called Nagel point. One can show that $M_{0}$ lies on the segment $I N$ and divides it in the ratio $1: 2$. Also note that each of the lines $A A_{2}, B B_{2}, C C_{2}$ divides the perimeter of the triangle in two equal parts. Our goal is to find analogues of the Euler line for an inscribed polygon and the Nagel line for a circumscribed one. 2.1. (A. Myakishev, II Sharygin olimpiad) Let $A B C D$ be an inscribed quadrilateral and $O$ be its circumcenter. Let $H_{a}, H_{b}, H_{c}, H_{d}$ be the orthocenters of triangles $B C D$, $C D A, D A B, A B C$ respectively, and $H$ be the intersection point of the lines $H_{a} H_{c}$ and $H_{b} H_{d}$. Show that the barycenter $M_{2}$ lies on the segment $O H$ and divides it in the ratio $1: 2$. 2.2. (A. Myakishev, II Sharygin olympiad) Let $A B C D$ be a circumscribed quadrilateral, and $I$ be its incenter. Let $T, U, V, W$ be the points which are symmetric to the points of tangency of the incircle with the sides $A B, B C, C D$, $D A$ respectively with respect to their midpoints. a) Show that any of lines $T V$ and $U W$ divides the perimeter of the quadrilateral into two equal parts. b) Let $N$ be the point of intersection of lines $T V$ and $U W$. Prove that $M_{2}$ lies on the segment $I N$ and divides it in the ratio $1: 2$. Another approach to the definition of the Euler line was proposed by I.Romanov $[4]$. Define the orthocenter of an inscribed $n$-gon $A_{1} \ldots A_{n}$ inductively. Let $H_{1}, \ldots, H_{n}$ be the orthocenters of ( $n-1$ )-gons $A_{2} \ldots A_{n}, \ldots, A_{1} \ldots A_{n-1}$ respectively. 2.3. Prove that the lines $A_{1} H_{1}, \ldots, A_{n} H_{n}$ are concurrent. 2.4. Let us call the corresponding intersection point $H$ the orthocenter of the $n$-gon. Show that the barycenter $M_{0}$ lies on the segment $O H$ and divides it in the ratio $(n-2): 2$. 2.5. Let $A B C D$ be an arbitrary quadrilateral. Consider two generalizations of the orthocenter: $H^{*}$ is the center of the parallelogram formed by the orthocenters of triangles $A B L, B C L, C D L, D A L$ $H^{* *}=H_{a} H_{c} \cap H_{b} H_{d}$, where as usual $H_{a}$ is the orthocenter of triangle $B C D$, and so on. Furthermore we generalize $O$ as $O^{* *}=O_{a} O_{c} \cap O_{b} O_{d}$ where $O_{a}$ is the circumcenter of triangle $B C D$, and so on (in other words, $O^{* *}$ is the intersection of perpendicular bisectors to $A C$ and $B D$ ). Prove that a) $M_{0}$ is the midpoint of $O^{* *} H^{*}$; b) (Ya. Ganin, A. Myakishev) $M_{2}$ lies on the segment $O^{* *} H^{* *}$ and divides it as $1: 2$; c) $H^{*}$ is the midpoint of $L H^{* *}$. ## 3 Quasi-centers of the circumcircle and the incircle In this section we will try to define the points $O$ and $I$ for an arbitrary quadrilateral, which have the properties similar to those of the circumcenter and the incenter. Of course, for an inscribed (resp. circumscribed) quadrilateral the point $O$ (resp. $I$ ) should coincide with the center of the circumcircle (resp. the incircle). 3.1. Show that for any quadrilateral $A B C D$ which is both inscribed and circumscribed, the centers $O, I$ and the intersection point $L$ of diagonals are collinear. 3.2. Let $I$ be the intersection point of the diagonals of a quadrilateral $P Q R S$. Denote the projections of $I$ to $P Q, Q R, R S, S P$ by $A, B, C, D$ respectively. Show that a) the quadrilateral $A B C D$ is circumscribed iff the quadrilateral $P Q R S$ is inscribed; b) if the quadrilateral $A B C D$ is circumscribed then $I$ is its incenter. Let $A^{\prime}, B^{\prime}, C^{\prime}, D^{\prime}$ be the intersection points of lines $I A, I B, I C, I D$ with $R S$, $S P, P Q, Q R$ respectively. 3.3. Prove that a) the quadrilateral $A B C D$ is inscribed iff $P R \perp Q S$; b) if $P R \perp Q S$, then $A^{\prime} B^{\prime} C^{\prime} D^{\prime}$ is a rectangle and the points $A, B, C, D, A^{\prime}, B^{\prime}$, $C^{\prime}, D^{\prime}$ are concyclic. 3.4. Construct a quadrilateral $P Q R S$ by the points $A, B, C, D$ if it is known that $I$ lies inside $A B C D$. Definition. Define the quasi-incenter of a convex quadrilateral $A B C D$ to be the point $I$ constructed in the previous problem, and the quasi-circumcenter to be the intersection point $O$ of the lines $A^{\prime} C^{\prime}$ and $B^{\prime} D^{\prime}$. (We assume that $I$ lies inside $A B C D$ ) 3.5. Show that the quasi-centers $O, I$ and the intersection point $L$ of the diagonals are collinear. 3.6. For a quadrilateral which is both inscribed and circumscribed express the circumradius and the inradius in terms of lengths of the segments $O I$ and $O L$. The last problem enables us to define the quasi-incircle and the quasi-circumcircle for an arbitrary quadrilateral. Up to date it is unknown whether these circles have any interesting properties. Now let us describe another approach to defining quasi-centers. 3.7. Let $I_{a}, I_{b}, I_{c}$ be the excenters of the triangle $A B C$, and $J$ be the circumcenter of $I_{a} I_{b} I_{c}$. Prove that $O$ is the midpoint of the segment $I J$. 3.8. Prove that for an arbitrary quadrilateral, its internal angle bisectors form an inscribed quadrilateral, and so do the external angle bisectors. Denote by $I$ and $J$ the circumcenters of the quadrilaterals formed by the internal angle bisectors and the external angle bisectors of $A B C D$ respectively. 3.9. (VII Sharygin olimpiad) Show that for an inscribed quadrilateral with circumcenter $O$, the points $I$ and $J$ are symmetric with respect to $O$. Now we can take $I$ and the midpoint $O$ of the segment $I J$ to be the quasi-incenter and the quasi-circumcenter. Unfortunately, with this definition the intersection point $L$ of the diagonals can be not contained in the line $O I$. One more approach to defining the quasi-circumcenter is proposed in [6]. 3.10. Let $X$ be the intersection point of lines $A B$ and $C D, Y$ be the intersection point of $A D$ and $B C, Z$ be the intersection point of $A C$ and $B D$. Let $M_{X}$ be the Miquel point of lines $A D, B C, A C$ and $B D, M_{Y}$ be the Miquel point of $A B, B D$, $A C$ и $B D$, and $M_{Z}$ be the Miquel point of $A D, B C, A B$ и $C D$. Prove that a) the lines $X M_{X}, Y M_{Y}$ and $Z M_{Z}$ are concurrent; b) if the points $A, B, C, D$ are concyclic then these lines intersect at the circumcenter of $A B C D$. The obtained point can also be considered as a quasi-circumcenter. ## 4 Additional problems 4.1. Let $A B C D$ be a quadrilateral without parallel sidelines circumscribed around a circle centered at $I$. The sides $A B, B C, C D, D A$ touche the incircle at points $X$, $Y, Z, T$ respectively. As usually $L=A C \cap B D$ (also $L=X Z \cap Y T$ ). Let $X^{\prime}$ be the reflection of $X$ about the midpoint $M_{A B}$ of side $A B ; Y^{\prime}, Z^{\prime}, T^{\prime}$ are defined similarly; $N=X^{\prime} Z^{\prime} \cap Y^{\prime} T^{\prime}$ is the Nagel point. Prove that the condition $M_{0}=I$ is equivalent to each of the following conditions: a) $A X+C Z=B Y+D T$ b) $X Z \| X^{\prime} Z^{\prime}$ (или $X Z \| M_{A B} M_{C D}$ ); c) $X^{\prime}, Z^{\prime}$ and $B C \cap A D$ are collinear; d) $L, I, N$ a collinear; e) (A.Zaslavsky, M.Isaev, D.Tsvetov, All-Russian olympiad 2005 г.) $I A \cdot I C=$ $I B \cdot I D$. 4.2. (A.Myakishev) Triangles $A B C$ and $A^{\prime} B^{\prime} C^{\prime}$ are called ortologic, if the perpendiculars from $A^{\prime}, B^{\prime}, C^{\prime}$ to $B C, C A, A B$ respectively concur. Quadrilaterals $A B C D$ and $A^{\prime} B^{\prime} C^{\prime} D^{\prime}$ are called ortologic, if the triangles $A B C$ and $A^{\prime} B^{\prime} C^{\prime}, B C D$ and $B^{\prime} C^{\prime} D^{\prime}, C D A$ and $C^{\prime} D^{\prime} A^{\prime}, D A B$ and $D^{\prime} A^{\prime} B^{\prime}$ are ortologic. Let $A B C D$ and $A^{\prime} B^{\prime} C^{\prime} D^{\prime}$ be ortologic, $A C$ and $B D$ meet at $L, A^{\prime} C^{\prime}$ and $B^{\prime} D^{\prime}$ meet at $L^{\prime}$. Prove that $A L: L C=A^{\prime} L^{\prime}: L^{\prime} C^{\prime}$ and $B L: L D=B^{\prime} L^{\prime}: L^{\prime} D^{\prime}$ (i.e. ortologic quadrilaterals are affine equivalent). ## Список литературы [1] http://faculty.evansville.edu/ck6/encyclopedia/ETC.html. [2] Ф.Ивлев. Центры тяжести многоугольников. Доклад на ММКШ. 2008. https://www.mccme.ru/circles/oim/mmks/notes.htm [3] A.Akopyan. Some remarks on the circumcenter of mass. https://arxiv.org/pdf/1512.08655.pdf [4] И.Романов. Прямая Эйлера n-угольника. Доклад на ММКШ. 2017. https://www.mccme.ru/circles/oim/mmks/works2017/ignatov2.pdf [5] А.Заславский. Диагонально-перпендикулярное отображение четырехугольников. Квант. 1998. No. 4. [6] M.Rolnek, Le Anh Dung. The Miquel Points, Pseudocircumcenter, and EulerPoncelet Point of a Complete Quadrilateral. Forum Geometricorum. V. 14 (2014). https://personal.us.es/rbarroso/trianguloscabri/sol/FG201413.pdf. ## Remarkable points of polygons
Solutions ## 1 Barycenters of polygons 1.1. Answer. The incenter of the triangle $A_{0} B_{0} C_{0}$. Proof. Note that the points $A_{0}, B_{0}, C_{0}$ are the midpoints of the segments $B C$, $C A, A B$, so let us place the weights equal to the lengths of these segments in this points. Then the barycenter of the weights in the points $A_{0}$ and $B_{0}$ is the point, which divides the segment $A_{0} B_{0}$ in the ratio $A C: B C=A_{0} C_{0}: B_{0} C_{0}$, i.e. the base of the angle bisector in $A_{0} B_{0} C_{0}$, therefore the barycenter of all three weights lies on this bisector. Analogously, we deduce that it lies on the other angle bisectors of $A_{0} B_{0} C_{0}$ too, hence it coincides with, its incenter (fig.1). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-078.jpg?height=656&width=885&top_left_y=1254&top_left_x=563) Fig. 1 1.2. It follows from the previous problem and the fact that the triangles $A B C$ and $A_{0} B_{0} C_{0}$ are homothetic with the center at $I$. 1.3. Denote by $X$ and $Y$ the points of tangency of excircles, which are tangent to the segments $A C$ and $B C$, with the line $A B$. We have $B X=A Y=p$ (where $p$ is the semiperimeter of the triangle), hence $C_{0} X=C_{0} Y$. Moreover, the line passing through the centers of these circles is orthogonal to the angle bisector of $C$, hence, it is also orthogonal to the line $C_{0} M_{1}$. THus we obtain that $M_{1}$ lies on the radical axis of these teo circles. Arguing in the same way we deduce that $M_{1}$ lies on the radical axis of any pair of the three excircles. 1.4. Let $X$ be the intersection of $C_{0} M_{1}$ with the segment $A C$. It follows from the previous problem that $X$ is the midpoint of the segment with the ends at the tangency points of the line $A C$ with the excircles, which are tangent to the segments $A C$ and $B C$ (fig.2). As the distances from these points to $A$ are equal to $p-c$ and $p$ respectively, we obtain that $A X=p-c / 2$ and $A X+A C_{0}=p$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-079.jpg?height=1068&width=1311&top_left_y=894&top_left_x=378) Fig. 2 1.5. Answer. $M_{0}$ is the center of the parallelogram $P Q R S$, where $P, Q, R$, Sare the midpoints of $A B, B C, C D, D A$. $M_{2}$ is the intersection point of the line $l_{1}$, passing through the centroids of $A B C$ and $A D C$, and the line $l_{2}$, passing through the centroids of $A B D$ and $B C D$. 1.6. Let $U, V$ be the midpoints of the diagonals $A C$ and $B D$ respectively and $L$ be their intersection point. The centroid of $A B C$ lies on its median $B U$ and divides it in the ratio $2: 1$. Analogously, the centroid of $A C D$ lies on $D U$ and divides it in the same ratio. The line $l_{1}$, passing through these centroids, is parallel to $B D$ and it intersects $A C$ at the point, which divides the segment $U L$ in the ratio $1: 2$. In the same way we obtain that $M_{2}$ lies on the line, which is parallel to $A C$ and which divides the segment $V L$ in the ratio $1: 2$. Note that the lines passing through $M_{0}$ parallel to $A C$ and $B D$ pass through the midpoints of the segments $U L$ and $V L$ respectively. The assertion of the problem now follows straightforwardly (fig.3). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-080.jpg?height=797&width=940&top_left_y=892&top_left_x=501) Fig. 3 1.7. As $P$ and $Q$ are the barycenters of the segments $A B$ and $B C$, we obtain that the barycenter $X_{1}$ of their union lies on $P Q$ and $P X_{1} / Q X_{1}=B C / A B$. This point mat be constructed in the following way: construct the angle bisector $B B^{\prime}$ of the triangle $B P Q$ and take a point, which is symmetric to $B^{\prime}$ with respect to the midpoint of $P Q$. Analogously one can construct the barycenters $X_{2}, Y_{1}, Y_{2}$ of the piecewise linear curves $C D A, D A B, B C D$. Then $M_{1}$ is the intersection point of the lines $X_{1} X_{2}$ and $Y_{1} Y_{2}$ (fig.4). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-081.jpg?height=760&width=965&top_left_y=471&top_left_x=474) Fig. 4 1.8 . a) Let $I$ be the incenter. The centroids of the (solid) triangles $I A B, I B C, I C D$, $I D A$ divide their medians $I P, I Q, I R, I S$ in the ratio $2: 1$, i.e. the quadrilateral, which is formed by them, is homothetical to $P Q R S$ with the center in $I$ and the coefficient $\frac{2}{3}$. As the ratio of the areas of $I A B, I B C, I C D, I D A$ is the same as the ratio of the corresponding sides of $A B C D$, this homothety maps $M_{1}$ to $M_{2}$. b) The proof is analogous to the previous one. 1.9. It follows from the previous problems and the theorem about three homothety centers. 1.10. Hint. $M_{0}$ is the midpoint of the segment, formed by the midpoints of the diagonals of the quadrilateral. Note that the Gauss line of the circumscribed quadrilateral contains its incenter. This fact implies that the locus of $M_{0}$ is a circle. Now one can use the corresponding homotheties to deduce that the loci of two other centers are also circles (fig.5). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-082.jpg?height=994&width=991&top_left_y=377&top_left_x=567) Fig. 5 ## 2 Euler and Nagel points 2.1. Let $M_{a}$ and $H_{a}$ be the centroid (the point $M_{2}$ ) and the orthocenter of $B C D$ respectively. Denote the centroids and the orthocenters of the other triangles analogously. All there triangle share a common circumcircle, whose center is $O$. By considering the Euler lines of these triangles, we note that the quadrilateral $M_{a} M_{b} M_{c} M_{d}$ is mapped to the quadrilateral $H_{a} H_{b} H_{c} H_{d}$ under the homothety with the center in $O$ and the coefficient 3 , so the intersection points of the corresponding diagonals are mapped to each other analogously. 2.2 . a) It follows from the definition that $A T+D V=B C, B T+C V=A D$, i.e. $T A+A D+D V=V C+C B+B T$. b) Let $a, b, c, d$ be the length of tangent segments to the incircle from the vertices $A, B, C, D$. It is clear that if one places the weights $a, b, c, d$ into the points $A, B$, $C, D$ respectively, then the barycenter of the obtained system is the point $N$, and if one places the weights $2 a+b+d, 2 b+a+c, 2 c+b+d, 2 d+c+a$ into those vertices, then the barycenter is the point $M_{1}$. It remains to show that $I$ is the barycenter of the weights $b+d, a+c, b+d, a+c$ and apply the assertion of problem 1.8.. The point $I$ satisfies $S_{I A B}-S_{I B C}+S_{I C D}-S_{I D A}=0$. The same property holds for the midpoints $U$ and $V$ of the diagonals of the quadrilateral, therefore these points are concurrent (this fact is called the Monge theorem). Now let $X$ and $Y$ be the tangency points of the incircle with the sides $B C$ and $A D$ respectively. Then the angles formed by the line $X Y$ and these sides are equal and also $X Y$ passes through the intersection point of the diagonals $L$ by Brianchon's theorem. Applying the laws of sines to the triangles $L X B$ and $L Y D$, we obtain that $B L / D L=b / d$. Similarly, $A L / C L=a / c$. These identities and the equalities $S_{U B C} / S_{U A D}=B L / D L$, $S_{V B C} / S_{V A D}=C L / A L, S_{I B C} / S_{I A D}=(b+c) /(a+d)$ imply that $I$ divides the segment $A C$ in the ratio $(a+c) /(b+d)$, QED. 2.3. Hint. Use induction. 2.4. Hint. Use induction. 2.5. a) Let $M_{A C}, M_{B D}$ be the midpoints of diagonals $A C, B D$. It is clear that $O^{* *} M_{A C} \perp A C$ and $O^{* *} M_{B D} \perp B D$. On the other hand $H^{*} M_{A C} \perp B D$ and $H^{*} M_{B D} \perp A C$ because $H^{*}$ is the center of the parallelogram having the sidelines perpendicular to the diagonals of the given quadrilateral and passing through its vertices. Hence $H^{*} M_{A C} O^{* *} M_{B D}$ is a parallelogram which yields the required assertion. b) Hint. Use the assertion of problem 4.2. c) Clearly follows from two previous assertions. ## 3 Quasi-centres of the circumcircle and the incircle 3.1. We use the following fact. Lemma. Suppose the point $P$ lies inside the circle with the center at $O$. Denote by $X$ and $Y$ the intersection points of the circle with two orthogonal half-lines with the initial point at $P$. Then the locus of the intersection points of the tangent lines to the circle at $X$ и $Y$ is a circle with the center, lying on $O P$. Proof. Let $Z$ be the fourth vertex of the rectangle $P X Z Y$. We have $O P^{2}+O Z^{2}=$ $O X^{2}+O Y^{2}$, thus the locus of $Z$ is the circle with the center at $O$. Hence the locus of the midpoint of the segment $X Y$ is a circle with the center at the midpoint of $O P$, so the locus of the intersection of the tangent lines, which is inverse to it, is also a circle. Now the assertion follows from the fact that the lines, passing through the tangency points of the opposite sides of the quadrilateral with the incircle, are orthogonal and they both pass through the intersection point of its diagonals. 3.2. As the quadrilaterals $I A Q B$ and $I B R C$ are inscribed, we have $\angle I B A=$ $\angle I Q A, \angle I B C=\angle I R C$, which imply both assertions. 3.3. The proof is similar to the previous problem. 3.4. Suppose that the half-lines $P Q$ and $S R$ intersect at $X$. Then we have $\angle P X S=\angle P I S-\angle I P A-\angle C S I$. The quadrilaterals IAPD and ICSD are inscribed, thus we obtain that $\angle I P A+\angle C S I=\angle C D A$. Besides, $\angle P I S=\angle R I Q=(\angle P A D+$ $\angle D C S+\angle R C B+\angle B A Q) / 2=(\angle A B C+\angle C D A) / 2$. Therefore, $\angle A I C=\pi-\angle P X S=$ $\pi-(\angle A B C-\angle C D A) / 2$. Arguing in the same way we can find the angle $B I D$ and thus to construct $I$ as the intersection point of the corresponding arcs of the circles, and then we are able to construct the quadrilateral (fig.6). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-084.jpg?height=625&width=1462&top_left_y=1555&top_left_x=434) Fig. 6 3.5. Hint. Apply the central projection, which maps the points $P, Q, R, S$ to the vertices of a parallelogram. 3.6. Answer. $R^{2}=\frac{O L \cdot O I^{2}}{2 O I-O L}, r^{2}=\frac{\left(R^{2}-O I^{2}\right)^{2}}{2\left(R^{2}+O I^{2}\right)}$. Hint. It follows from Poncelet theorem that it is enough to consider the quadrilateral having a diameter of the circumcircle as one of the diagonals. 3.7. Hint. Prove that the segments $I_{a} J, I_{b} J, I_{c} J$ are orthogonal to the corresponding sides of the triangle $A B C$. 3.8. One can prove this by a direct computation of angles. 3.9. Let the angle bisectors of $A$ and $B$ intersect at $K$, the angle bisectors of $B$ and $C$ - at $L$, the angle bisectors of $C$ and $D-$ at $M$, the angle bisectors of $D$ and $A$ - at $N$ (pic.9.4). Then the line $K M$ is the bisector of the angle, formed by $A D$ and $B C$. Denote this angle by $\phi$, exterior angle theorem implies that $\angle L K M=\angle B / 2-\phi / 2=(\pi-\angle A) / 2=\angle C / 2$, hence $\angle L I M=\angle C$. On the other side, the line through $L$ orthogonal to $B C$ and the line through $M$ orthogonal to $C D$ form with $M L$ the angles, equal to $(\pi-\angle C) / 2$, i.e. the triangle, formed by these lines and $M L$ is isosceles and its vertex angle is equal to $C$. Thus its vertex coincides with $I$. Therefore the lines through the vertices of $K L M N$ orthogonal to the respective sides of $A B C D$ pass through $I$ (fig.7). Similarly we obtain that the perpendiculars from the vertices of the quadrilateral formed by the exterior angle bisectors pass through $J$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-085.jpg?height=814&width=825&top_left_y=1443&top_left_x=638) Fig. 7 Now let $K^{\prime}$ be the intersection points of the exterior bisectors of $A$ and $B$. As the quadrilateral $A K B K^{\prime}$ is inscribed in a circle with the diameter $K K^{\prime}$, we obtain that the projections of $K$ and $K^{\prime}$ to $A B$ are symmetric with respect to the midpoint of $A B$. From this assertion and the one, proved above, it follows that the projections of $I$ and $J$ to each of the sides of $A B C D$ are symmetric with respect to the midpoints of the corresponding sides, which is equivalent to the assertion of the problem. 3.10. See [6]. ## Список литературы [1] http://faculty.evansville.edu/ck6/encyclopedia/ETC.html. [2] Ф.Ивлев. Центры тяжести многоугольников. Доклад на ММКШ. 2008. https://www.mccme.ru/circles/oim/mmks/notes.htm [3] A.Akopyan. Some remarks on the circumcenter of mass. https://arxiv.org/pdf/1512.08655.pdf [4] И.Романов. Прямая Эйлера $n$-угольника. Доклад на ММКШ. 2017. https://www.mccme.ru/circles/oim/mmks/works2017/ignatov2.pdf [5] А.Заславский. Диагонально-перпендикулярное отображение четырехугольников. Квант. 1998. №4. [6] M.Rolnek, Le Anh Dung. The Miquel Points, Pseudocircumcenter, and EulerPoncelet Point of a Complete Quadrilateral. Forum Geometricorum V. 14 (2014). https://personal.us.es/rbarroso/trianguloscabri/sol/FG201413.pdf. # Вокруг теоремы Кэли Задачу представляют Бурсиан О., Кохась Д., Кохась К. Версия 0.99 Этот проект посвящен теореме Кэли о числе деревьев. Дерево - это связный граф без циклов. Теорема утверждает, что число помеченных деревьев с $n$ вершинами равно $n^{n-2}$. Известно много доказательств этой теоремы, нашей целью будет познакомиться с некоторыми из них и изучить приложения различных подходов к этой теореме. В каждом проекте есть довольно сложные задачи. По правилам конференции задачи можно решать, объединившись в команды; задачи разных проектов можно решать в составе разных команд. Мы советуем вам найти попутчиков, с которыми вам было бы интересно проводить исследования, описанные в этом проекте. ## Задачи для первого знакомства В олимпиадных задачах обычно используются графы с «обезличенными» вершинами - например, вершины - это какие-то города, а ребра - дороги между ними, или вершины - это люди (обычно безымянные), а ребра - это знакомства и т.п. В задачах о перечислении графов принята другая точка зрения - все вершины графа должны быть «индивидуальны». Чтобы не было диссонанса в терминологии, мы введем понятие «помеченный граф». Пусть $[n]=\{1,2, \ldots, n\}$. Дерево (или вообще произвольный граф) на $n$ вершинах, вершины которого пронумерованы числами от 1 до $n$, называется помеченным деревом (соответственно помеченным графом). Чтобы построить помеченное дерево, можно взять дерево и расставить в его вершинах числа, или наоборот: мы можем считать, что множество $[n]$ - это множество вершин, и мы рисуем граф-дерево, соединяя эти вершины ребрами. Множество помеченных деревьев на $n$ вершинах обозначим через $\mathcal{T}_{n}$. Два (непомеченных) графа $G_{1}$ и $G_{2}$ с множеством вершин $V_{1}$ и $V_{2}$ называются изоморфными (или, попросту говоря, считаются одинаковыми), если существует такое взаимно однозначное отображение $f: V_{1} \rightarrow V_{2}$, что вершины $A, B \in V_{1}$ соединены ребром в $G_{1}$ тогда и только тогда, когда $f(A)$ и $f(B)$ соединены ребром в $G_{2}$. Например, любое дерево с четырьмя вершинами изоморфно либо дереву «куриная лапа», либо дереву «путь длины три». В случае, когда $G_{1}$ и $G_{2}$ - помеченные графы, $V_{1}=V_{2}=[n]$ и в качестве $f$ в этом определении берут тождественное отображение. Мы предполагаем, что при решении задач разделе «Задачи для первого знакомства» теорема Кэли не должна использоваться. В формулировках задач через $T_{n}$ обозначается количество помеченных деревьев на $n$ вершинах, и вопрос о нахождении этой величины не ставится. Нумерация задач дана в соответствии с последующими разделами. 1.1. Граф называется унициклическим, если он связный и содержит ровно один цикл. Докажите, что помеченных деревьев со 100 вершинами больше, чем помеченных унициклических графов с 98 вершинами. 1.2. Постройте взаимно однозначное соответствие между множеством, которое состоит из всевозможных отображений множества $[n]$ в себя, и множеством помеченных деревьев на $n$ вершинах, в которых одна вершина снабжена красной меткой и одна - синей меткой (может оказаться, что обе метки поставлены возле одной и той же вершины). 1.3. Существует $n^{n-1}$ различных отображений из множества $[n-1]$ в $[n]$. Докажите тождество $$ \sum_{j=1}^{n} C_{n-1}^{j-1}(n-j)^{n-j} T_{j}=n^{n-1} $$ разбив множество этих отображений на $n$ частей так, чтобы $j$-е слагаемое оказалось равно количеству отображений в $j$-й части. 3.1. Дерево с $n$ вершинами, ребра которого пронумерованы числами от 1 до $n-1$, называется реберно помеченным. Например, существует 4 различных реберно помеченных дерева на 4 вершинах - см. рис. 1. Докажите, что при $n \geqslant 3$ количество различных реберно помеченных деревьев на $n$ вершинах равно $\frac{1}{n} T_{n}$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-089.jpg?height=142&width=958&top_left_y=414&top_left_x=548) Рис. 1. Реберно помеченные курлапа и пути длины 3 Если мы выделили в дереве (помеченном или нет) одну из вершин, то будем называть такое дерево корневым, а саму отмеченную вершину будем называть корнем. Если же нам потребуется подчеркнуть, что ни одна из вершин нами не выделена, будем называть такое дерево свободным. Листом дерева называется вершина степени 1 , за исключением случая, когда дерево корневое и корень имеет степень 1 , в этом случае корень листом не считается. Лес - это граф, в котором каждая компонента связности является деревом. Обозначим через $\mathcal{F}_{n}^{k}$ множество лесов на множестве вершин $[n]$, состоящих из $k$ корневых деревьев с корнями $1,2, \ldots, k$, причем таких, у которых вершина $n$ находится в дереве с корнем 1 (рис. 2). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-089.jpg?height=446&width=945&top_left_y=1022&top_left_x=561) Рис. 2. Множество лесов $\mathcal{F}_{5}^{3}$. Мы ориентировали ребра «от периферии к корню» Множества мы обозначаем «рукописными» буквами. Число элементов множества обозначается той же буквой в курсивном начертании. Так число в множестве $\mathcal{F}_{n}^{k}$ будем обозначать $F_{n}^{k}$. 1.4. Докажите, что при $2 \leqslant k \leqslant n-1$ выполняется рекуррентное соотношение $$ F_{n}^{k-1}=n F_{n}^{k} $$ 1.5. Пусть $V_{1}=[r], V_{2}=\{r+1, \ldots, r+s\}, V=V_{1} \cup V_{2}=[r+s]$. Обозначим через $\mathcal{F}_{r, s}^{k}$ множество лесов, состоящих из $k$ корневых деревьев с множеством вершин $V$ с корнями 1 , $2, \ldots, k$, у которых вершина $r+1$ находится в дереве с корнем 1 , а у каждого ребра одна вершина из множества $V_{1}$, а другая - из множества $V_{2}$. Придумайте рекурсию (по $k$ ) для величины $F_{r, s}^{k}$ Вдоль улицы с односторонним движением расположено $n$ мест для парковки автомобиля. На улицу въезжают последовательно $n$ автомобилей с номерами от 1 до $n$ в порядке возрастания. Каждый водитель едет к своему любимому месту парковки и, если оно не занято, припарковывается; в противном случае он едет дальше до первого свободного места и припарковывается там, если же все места дальше заняты, он уезжает (насовсем). Последовательностью предпочтений назовем список $a_{1}, a_{2}, \ldots, a_{n}$ любимых мест парковки первого, второго, $\ldots, n$-го водителя. 4.1. Докажите, что существует $(n+1)^{n-1}$ последовательностей предпочтений, для которых все водители сумеют припарковаться. 4.2. Докажите, что успех или неуспех парковки на самом деле не зависит от порядка, в котором прибывают машины. Следующая задача лежит в стороне от нашего сюжета. Но она позволяет понять, какие затруднения могут возникнуть, если мы станем перечислять непомеченные деревья. 3.2. Докажите, что количество различных (т. е. неизоморфных друг другу) непомеченных деревьев с $n$ вершинами меньше $4^{n}$. ## 1 Рекурсии, тождества, биекции 1.6. Докажите при помощи комбинаторных рассуждений (интерпретируя числа $T_{i}$ как количества деревьев), что при $n>1$ а) $T_{n}=\frac{n}{2} \sum_{k=0}^{n-2} C_{n-2}^{k} T_{k+1} T_{n-k-1}$; b) Сведите эту формулу, а также формулы из задач 1.3 и 5.1 a) друг к другу алгебраически. 1.7. Обозначим через $\mathcal{T}(n, k)$ множество помеченных корневых деревьев с $n$ вершинами, в которых корень - вершина 1 - имеет степень $k$. Докажите, что $$ (n-1)(k-1) T(n, k)=(n-k) T(n, k-1) $$[^4] ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-090.jpg?height=160&width=1614&top_left_y=1094&top_left_x=224) Рис. 3. Непомеченные треугольные деревья на $2,3,4,5$ и 6 вершинах. Например, для $n=5$ имеется 2 непомеченных треугольных дерева (рис. 3) и 70 помеченных. Если некоторое ребро треугольного дерева выделено, то такое дерево будем называть «корневым». 1.8. Обозначим через $\Delta(n, k)$ число корневых помеченных треугольных деревьев на $n$ вершинах, у которых корневое ребро принадлежит $k$ треугольникам. Придумайте рекурсию (по $k$ ) для величины $\Delta(n, k)$. ## 2 Код Прюфера Код Прюфера сопоставляет дереву с занумерованными вершинами последовательность его вершин следующим образом. Код Прюфера дерева с двумя вершинами - пустое слово. Если количество вершин дерева $T$ больше двух, то обозначим через $v$ лист с минимальным номером, а через $u$ вершину, смежную с $v$. Тогда код Прюфера дерева $T$ получается из кода Прюфера дерева $T-v$ приписыванием слева вершины $u$. Проверьте, что Вы умеете решать задачу 2.1 , и зарегистрируйте плюсик у жюри. 2.1. а) Найдите код Прюфера дерева с вершинами $1,2, \ldots, 10$ и рёбрами $(8,9),(8,4),(4,10)$, $(10,3),(3,5),(10,6),(10,1),(1,7),(1,2)$. b) Восстановите дерево по коду Прюфера 1, 1, 2, 5, 4, 2, 7 . c) Докажите, что код Прюфера определяет взаимно однозначное соответствие между множеством деревьев с данными $n$ вершинами и множеством слов длины $n-2$ из этих вершин. d) Докажите, что в коде Прюфера вершина степени $d$ встречается $d$ - 1 раз. 2.2. Чему равно количество корневых помеченных деревьев с $n$ вершинами, у которых вершина $n$ является листом (и тем самым она не корень)? 2.3. Чему равно количество свободных помеченных деревьев с $n>10$ вершинами, у которых степень вершины 1 равна 10 ? 2.4. Найдите количество помеченных унициклических графов с $n$ вершинами, имеющих цикл длины $k$. 2.5. Обозначим через $S(n, k)$ число Стирлинга второго рода, по определению оно равно количеству разбиений множества $[n]$ на $k$ непустых частей. Докажите, что количество помеченных деревьев с $n$ вершинами, ровно $r$ из которых - листья, равно $\frac{n!}{r!} S(n-2, n-r)$. 2.6. Обозначим через $\tau\left(k_{1}, k_{2}, \ldots, k_{n}\right)$ число помеченных свободных деревьев с $n$ вершинами, у которых степень $i$-й вершины равна $k_{i}+1$. Докажите, что $\tau\left(k_{1}, k_{2}, \ldots, k_{n}\right)=\frac{(n-2)!}{k_{1}!k_{2}!\ldots k_{n}!}$. ## 3 Резулътати В следующих задачах можно сдавать несколько решений, если они принципиально отличаются. 3.3. Выведите теорему Кэли из задач: а) 1.4 , b) 1.6 , c) 1.7 , d) 2.6 . 3.4. Докажите, что количество лесов, с вершинами в множестве $[n]$, состоящих из $k$ корневых деревьев, равно $C_{n-1}^{k-1} n^{n-k}$. 3.5. Докажите, что количество лесов, с вершинами в множестве $[n]$, состоящих из двух некорневых деревьев, равно $\frac{1}{2} n^{n-4}(n-1)(n-6)$. 3.6. Докажите, что количество остовных деревьев в полном двудольном помеченном графе $K_{r, s}$ с долями $V_{1}=[r]$ и $V_{2}=\{r+1, \ldots, r+s\}$ равно $r^{s-1} s^{r-1}$. 3.7. Пусть $\Delta_{n}$ - число помеченных треугольных деревьев с $n$ вершинами, $\Lambda_{n}$ - число корневых помеченных треугольных деревьев с $n$ вершинами и корневым ребром $1-2$. a) Докажите, что $\Lambda_{n}=(2 n-3)^{n-3}$. b) Найдите $\Delta_{n}$. Пусть в ряд расположено $n$ различных предметов. Циклическая перестановка - это перестановка, перемещающая предметы по некоторому циклу: один предмет ставится на место другого, этот другой - на место третьего, и т. д., последний ставится на место первого. Транспозиция ( $i j$ ) - это перестановка, меняющая местами предметы на $i$-м и $j$-м месте. Если считать, что множество $[n]$ является множеством вершин некоторого графа, то транспозицию ( $i j$ ) можно интерпретировать как ребро, соединяющее вершину $i$ с вершиной $j$. Результат последовательного выполнения перестановок $s_{1}, s_{2}, \ldots, s_{n-1}$ определяет перестановку, которая называется произведением этих перестановок и обозначается $s_{1} s_{2} \ldots s_{n-1}$. Произведения, отличающиеся порядком сомножителей, мы считаем различными. Например, если перестановка $s$ - это транспозиция, меняющая местами предметы на первом и втором месте, а перестановка $t$ - это транспозиция, меняющая местами предметы на третьем и четвертом месте, то произведения $s t$ и $t s$ определяют одну и ту же перестановку предметов, и при этом считаются разными произведениями. 3.8. а) В ряд расположено $n$ предметов. Докажите, что результат последовательного выполнения транспозиций $s_{1}, s_{2}, \ldots, s_{n-1}$ является циклической перестановкой в том и только том случае, когда граф, в котором множество вершин - это $[n]$, а множество ребер - это $s_{1}, s_{2}, \ldots, s_{n-1}$, является деревом. b) Докажите, что количество способов, которым циклическая перестановка множества $[n]$ может быть разложена в произведение $n-1$ транспозиций, равно $T_{n}$. ## 4 Парковочные функции Вдоль улицы с односторонним движением расположено $n$ мест для парковки автомобиля. На улицу въезжают последовательно $n$ автомобилей с номерами от 1 до $n$ в порядке возрастания. Каждый водитель едет к своему любимому месту парковки и, если оно не занято, припарковывается; в противном случае он едет дальше до первого свободного места и припарковывается там, если же все места дальше заняты, он уезжает (насовсем). Последовательность предпочтений, для которой все водители сумели припарковаться, называется парковочной функцией. Множество парковочных функций обозначим через $\mathcal{P}_{n}$. 4.3. Сколько существует парковочных функции $\left(a_{1}, a_{2}, \ldots, a_{n}\right)$, для которых соседние водители имеют разные предпочтения, т. е. $a_{k} \neq a_{k+1}$ при $k=1, \ldots, n-1$ ? 4.4. Допустим, что на улицу, где расположено $n$ мест для парковки, въезжают лишь $ma_{2}>\ldots>$ $a_{k}$ - текущие пометки её и всех её потомков в убывающем порядке (тогда $a_{1}$ - пометка в $v$, и $k>i$ ). Переставим метки $a_{1}, \ldots, a_{i+1}$ так: поставим $a_{i+1}$ в $v$, а $a_{j+1}$ заменим на $a_{j}$ при всех $j=1,2, \ldots, i$. Нетрудно понять, что инверсность $v$ стала равна $i$, никакие два потомка $v$ инверсии не образуют, и инверсности остальных вершин не поменялись (ибо не поменялись множества их потомков). Значит, обработав таким образом все вершины, мы получим дерево $T$ с требуемыми свойствами. На рисунке ниже показан пример действия этого алгоритма (индексы указывают плохости вершин). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-112.jpg?height=474&width=1336&top_left_y=1979&top_left_x=360) Рис. 7. Построение $T$ по $E$ (и наоборот) Чтобы построить обратное отображение из $\mathcal{T}_{n+1}$ в $\mathcal{E}_{n+1}$, заметим, что плохости вершин в $E$ расставляются автоматически, исходя из условия (b). Перестановка же пометок строится похожим образом снизу вверх. Когда обрабатывается вершина $v$, никакие два её потомка не образуют инверсии. Пусть $a_{1}>\ldots>a_{k}$ - пометки её и её потомков, причём в $v$ стоит $a_{i+1}$. Тогда достаточно переставить метки по циклу $a_{1} \rightarrow a_{i+1} \rightarrow a_{i} \rightarrow \ldots \rightarrow a_{1}$. В результате этого процесса получится требуемое о-дерево. Легко видеть, что построенные отображения взаимно обратны. (ii): Биекция между $\mathcal{P}_{n}$ и $\mathcal{E}_{n+1}$. Пусть $a=\left(a_{1}, \ldots, a_{n}\right)$ - парковочная функция, а $p_{1}, \ldots, p_{n}$ - номера мест, на которых припарковались соответствующие водители; тогда $p=\left(p_{1}, \ldots, p_{n}\right)-$ перестановка чисел $1,2, \ldots, n$. Назовём неудобством $i$-го водителя величину $p_{i}-a_{i}$; суммарное неудобство всех водителей есть неудобство функции $a$. Мы устроим биекцию следующим образом. По функции а мы построим о-дерево $E \in \mathcal{E}_{n+1}$ такое, что (c) плохость вершины с пометкой $k$ равна неудобству водителя $k$. Как следствие, $\operatorname{bad}(E)=D(a)$, а значит, композиция двух наших биекций - требуемая. Осталось, собственно, предъявить вторую биекцию. Мы начнём с более простой биекции - между перестановками $p=\left(p_{1}, \ldots, p_{n}\right)$ и помеченными деревьями $T \in \mathcal{T}_{n+1}$ без инверсий. Затем мы продолжим эту биекцию на парковочные функции и оснащённые деревья. Наг 1. Сопоставление перестановок деревъям без инверсий. Положим $p_{n+1}=n+1$ и рассмотрим числа $1,2, \ldots, n+1$ как вершины строящегося дерева; назовём число $p_{i}$ местом вершины $i$. Для каждой вершины $i \leqslant n$ найдём наименьшее из мест $p_{j}$ при $j>i$, которое больше $p_{i}$, и соединим $i$ с $j$. Мы получим связный граф (все вершины соединены путём с $n+1$ ) с $n$ рёбрами, т. е. дерево $T$. (Процесс получения этого дерева удобно представлять себе «справа налево»: мы по очереди рассматриваем вершины $n, n-1, \ldots, 1$ и подсоединяем их к получающемуся справа дереву согласно правилу.) Исследуем полученное дерево. Ясно, что в описанной выше ситуации $i-$ сын $j$. Выберем произвольную вершину $j \leqslant n$. Для любого потомка $k$ вершины $j$ выполнены неравенства $k<$ $j$ и $p_{k}j-$ вершина, для которой $p_{s}i$ - та вершина, к которой $i$ подсоединится в обратной процедуре. Тогда $p_{i}p_{s}$, так что $p_{j}-a_{j} \leqslant p_{j}-p_{s}-1$, т. е. плохость вершины $j$ не превосходит числа её потомков. Итого, мы получили о-дерево $E$. Наоборот, по о-дереву $E$ (точнее, по подлежащему помеченному дереву $T$ ) можно восстановить перестановку $p$, а по ней - функцию $a$, опять же согласно (c). Ясно, что если по $a$ псстроить $E$, то по $E$ восстановится именно функция $a$. Осталось понять, что, наоборот, если по произвольному $E \in \mathcal{E}_{n+1}$ построить функцию $a$, то по ней обратно построится дерево $E$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-114.jpg?height=403&width=1263&top_left_y=1129&top_left_x=405) Рис. 9. Оснащение дерева Для начала покажем, что полученная функция $a$ - парковочная. Поскольку $a_{i} \leqslant p_{i}$, а $\left(p_{i}\right)$ - перестановка, для этого достаточно показать, что $a_{i}>0$. Это просто: если у вершины $i$ есть $d$ потомков, все они имеют номера, меньшие $p_{i}$; значит, $p_{i}$ больше, чем $d$ - и, как следствие, чем плохость вершины $j$. Отсюда и следует требуемое. Осталось показать, что, если по о-дереву $E$ получены перестановка $p=\left(p_{i}\right)$ и парковочная функция $a$, то, действуя по этой парковочной функции, водители образуют ровно перестановку $p$. Для начала вспомним, что согласно (d) места $p_{j}$ всех потомков $j$ вершины $i$ и её самой образуют множество $A_{j}=\left\{p_{i}-d, \ldots, p_{j}-1, p_{j}\right\}$, где $d$ - число потомков $j$. Значит, $a_{j} \in A_{j}$. Теперь нетрудно показать, что водитель $i$ попадёт на место $p_{i}$, индукцией по $i$. При $i=1$ это верно, ибо 1 - лист нашего дерева, то есть $p_{1}=a_{1}$. Пусть все водители $j1$ а) $T_{n}=\frac{n}{2} \sum_{k=0}^{n-2}\binom{n-2}{k} T_{k+1} T_{n-k-1}$; b) Reduce this formula and also the formulas from problems 1.3 and 5.1 a) to each other algebraically. 1.7. Denote by $\mathcal{T}(n, k)$ the set of labelled rooted trees with $n$ vertices in which the root is labelled by 1 and has degree $k$. Prove that $$ (n-1)(k-1) T(n, k)=(n-k) T(n, k-1) $$ #### Abstract A "triangle tree" is a graph defined by induction in the following way. The smallest triangle tree is a complete graph with two vertices (in contrast with its name). If some triangle tree is already given we can take its arbitrary edge $A B$, take new vertex $C$ and add vertex $C$ and edges $A C, B C$ to the tree. By a labelled triangle tree we call a triangle tree with vertices numbered from 1 to $n$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-118.jpg?height=169&width=1629&top_left_y=1092&top_left_x=222) Figure 3. Unlabelled triangle trees on $2,3,4,5$ and 6 vertices For example, for $n=5$ there exist 2 unlabelled and 70 labelled triangle trees (fig. 3). If some edge of a triangle tree is singled out then we call such tree "rooted". 1.8. Denote by $\Delta(n, k)$ the number of rooted labelled triangle trees with $n$ vertices such that its rooted edge belongs to $k$ triangles. Find recursion (by $k$ ) for the numbers $\Delta(n, k)$. ## 2 Prüfer code Prüfer code corresponds a tree with numbered vertices to the sequence of its vertices in the following way. Priffer code of a tree with two vertices is an empty word. If the number of vertices of a tree $T$ is more than 2 then denote by $v$ a leaf with smallest number, and by $u$ the vertex adjacent to $v$. Then Prüfer code of tree $T$ is obtained from Prüfer code of tree $T-v$ by appending the vertex $u$ (to the left). Check that you know how to solve problem 2.1 and come to jury to register your plus. 2.1. a) Find Prüfer code of tree with vertices $1,2, \ldots, 10$ and edges $(8,9),(8,4),(4,10),(10,3)$, $(3,5),(10,6),(10,1),(1,7),(1,2)$. b) Reconstruct tree by the Prüfer code 1, 1, 2, 5, 4, 2, 7 . c) Prove that Prüfer code defines one-to-one correspondence between the set of trees on the given set of $n$ vertices and the set of words of length $n-2$ with "letters" from this set. d) Prove that a vertex of degree $d$ occurs $d-1$ times in the Prüfer code. 2.2. What is the number of rooted labelled trees with $n$ vertices in which vertex $n$ is a leaf (and hence it is not root)? 2.3. What is the number of free labelled trees with $n>10$ vertices in which the degree of vertex 1 is equal to 10 ? 2.4. Find the number of labelled unicyclic graphs with $n$ vertices having a cycle of length $k$. 2.5. Denote by $S(n, k)$ Stirling number of the second kind, by definition it is equal to the number of ways to partition set $[n]$ into $k$ nonempty parts. Prove that the number of labelled trees on $n$ vertices with exactly $r$ leafs is equal to $\frac{n!}{r!} S(n-2, n-r)$. 2.6. Denote by $\tau\left(k_{1}, k_{2}, \ldots, k_{n}\right)$ the number of free labelled trees with $n$ vertices such that the degree of $i$-th vertex is $k_{i}+1$. Prove that $\tau\left(k_{1}, k_{2}, \ldots, k_{n}\right)=\frac{(n-2)!}{k_{1}!k_{2}!\ldots k_{n}!}$. ## 3 Results You can submit several solutions of the next problems if they are quite different. 3.3. Deduce Cayley's theorem from problems: a) 1.4, b) 1.6 , c) 1.7 , d) 2.6 . 3.4. Prove that the number of forests with vertices from set $[n]$ consisting of $k$ rooted trees is equal to $\binom{n-1}{k-1} n^{n-k}$. 3.5. Prove that the number of forests with vertices from set $[n]$ consisting of two free trees is equal to $\frac{1}{2} n^{n-4}(n-1)(n-6)$. 3.6. Prove that the number of spanning trees of complete bipartite labelled graph $K_{r, s}$ with parts $V_{1}=[r]$ and $V_{2}=\{r+1, \ldots, r+s\}$ is equal to $r^{s-1} s^{r-1}$. 3.7. Let $\Delta_{n}$ be the number of labelled triangle trees with $n$ vertices, $\Lambda_{n}$ be the number of rooted labelled triangle trees with $n$ vertices and rooted edge 1-2. a) Prove that $\Lambda_{n}=(2 n-3)^{n-3}$. b) Find $\Delta_{n}$. Let $n$ different objects are arranged in the row. A cyclic permutation is a permutation that moves all the objects along some cycle: it puts the first object on the place of the second one, puts the second one on the place of the third one and so on, the last object is put on the place of the first one. A transposition (ij) is a permutation that changes objects on the $i$-th и $j$-th place. If we consider $[n]$ as set of vertices of some graph then the transposition $(i j)$ can be interpreted as an edge connecting vertex $i$ with vertex $j$. The result of consecutive applying of permutations $s_{1}, s_{2}, \ldots, s_{n-1}$ is a permutation that is called a product of these permutations and is denoted by $s_{1} s_{2} \ldots s_{n-1}$. We consider the products that differ by the order of factors as different. For example, if permutation $s$ is a transposition changing objects on the first and second places, permutation $t$ is a transposition changing objects on the third and fourth places, then the products $s t$ and $t s$ determine the same permutations but we consider them as different products. 3.8. a) $n$ objects are arranged in the row. Prove that the result of consecutive applying of transpositions $s_{1}, s_{2}, \ldots, s_{n-1}$ is a cyclic permutation if and only if the graph with vertex set $[n]$ and edge set $s_{1}, s_{2}, \ldots, s_{n-1}$ is a tree. b) Prove that the number of ways in which a cyclic permutation of set $[n]$ can be represented as a product of $n-1$ transpositions is $T_{n}$. ## 4 Parking functions There are $n$ parking spaces available along a one-way street and each of $n$ drivers numbered from 1 to $n$ has a preferred parking space. The drivers arrive consecutively in increasing order of their numbers. Each driver goes to his preferred parking place and parks on that place if it's not occupied; otherwise he goes farther to first free space and parks there, if all the spaces are occupied he goes away forever. A preference sequence for which everyone finds a parking space is called a parking function. The set of parking functions is denoted by $\mathcal{P}_{n}$. 4.3. Find the number of parking functions $\left(a_{1}, a_{2}, \ldots, a_{n}\right)$ such that any two consecutive drivers have different preferences that is $a_{k} \neq a_{k+1}$ for $k=1, \ldots, n-1$ ? 4.4. Let only $m1$ из $2(m+n)-4$ граничных клеток прямоугольника $m \times n$. Назовём раскраску красивой, если единицы стоят на объединении какого-то (конечного или бесконечного) множества рамок, а в остальных клетках стоят нули. Пусть задано конечное множество локальных правил, и все красивые раскраски клетчатой плоскости являются разрешёнными. Докажите, что есть бесконечно много неэквивалентных разрешённых раскрасок, не являющихся красивыми. B8 Пусть заданы некоторые локальные правила и, в соответствии с этими правилами, для любого положительного $r$ плитками можно замостить область, включающую круг радиуса $r$. Докажите, что тогда можно замостить и всю плоскость. С помощью локальных правил можно получать довольно сложные замощения. Например, с помощью них можно добиться того, чтобы все разрешенные замощения были непериодичны. Сразу это задачу будет решить сложно, мы рекомендуем вернуться к ней в цикле $C$, после освоения дополнительных методов. B9* Придумайте локальные правила такие, чтобы все разрешенные замощения были непериодичны. B10* Постройте конечный набор многоугольников, которыми можно замостить плоскость только непериодически. Многоугольники друг к другу можно прикладывать как угодно, можно поворачивать и переворачивать, но они не должны перекрываться. Определение. Пусть каждому символу алфавита сопоставлен квадрат $k \times k$ из символов того же алфавита. Такое соответствие будем называть подстановкой. Если есть подстановка $\sigma$ и замощение $A$ какой-то области (конечной или бесконечной), то можно получить новое замощение $\sigma(A)$. Для этого каждую плитку мы заменяем на соответствующий квадрат $k \times k$. Если начать с одной плитки и применять подстановку несколько раз, мы получим большой квадрат. Определение. Пусть задана некоторая $k \times k$ подстановка $\sigma$. Пусть также замощение плоскости можно разбить вертикальными и горизонтальными прямыми на квадраты $k \times k$ так, что каждый квадрат $k \times k$ лежит в образе $\sigma$ (то есть каждый квадрат является $\sigma(A)$ для некоторой буквы $A$ ). Запишем вместо каждого квадрата $\sigma(A)$ букву $A$. Полученное замощение будем обозначать как $\sigma^{-1}(S)$. Если $\sigma^{-1}(S)$ определено однозначно, будем говорить, что в замощении $S$ возможен переход к прообразу $\sigma^{-1}(S)$. Замощение будем называть бесконечно декодируемым относительно подстановки $\sigma$, если переход к прообразу можно провести любое число раз. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-141.jpg?height=201&width=786&top_left_y=1737&top_left_x=716) B11 Рассмотрим подстановку как на рисунке. Бесконечно декодируемое замощение для неё называется ковром Серпинского. a) Докажите, что есть бесконечно много попарно неэквивалентных ковров Серпинского. b) Докажите, что все ковры Серпинского кроме одного являются непериодичными замощениями. c) Задаётся ли множество ковров Серпинского локальными правилами? Если мы хотим задать с помощью локальных правил бесконечно декодируемые замощения, логично потребовать, чтобы подстановка переводила разрешённые замощения в разрешённые. Определение. Будем говорить, что подстановка согласована с локальными правилами размера $k$, если 1) для любой буквы $A$, квадрат $\sigma(A)$ является разрешенным; 2) если $X$ - разрешенный квадрат $2 \times 2$, то квадрат $\sigma(X)$ содержит только разрешенные квадраты $k \times k$. В12 Пусть заданы локальные правила $R$, причем $k=2$, то есть все запрещенные квадраты размера $2 \times 2$. Пусть некоторый квадрат $A$ размера $N \times N$ является разрешенным, то есть не содержит запрещенных квадратов $2 \times 2$. Пусть подстановка $\sigma$ согласована с $R$. Докажите, что квадрат $\sigma(A)$ также разрешенный.[^5] B13 Пусть подстановка $\sigma$ согласована с локальными правилами размера 2. Докажите, что существует разрешенное замощение. B14 Рассмотрим подстановку $\sigma$. Верно ли, что существует замощение $S$, для которого верно $\sigma(S)$ эквивалентно $S$ ? Придумайте достаточные условия на $\sigma$, чтобы такое замощение существовало. ## С. ПРИМЕРЫ НЕПЕРИОДИЧнОСТИ Итоги после циклов А и В. Мы хотим найти такое множество замощений, которое с одной стороны задаётся локальными правилами, а с другой стороны - содержит только непериодичные замощения. В одномерном случае такого множества не бывает (почему?). Самые простые известные нам двумерные примеры это бесконечно декодируемые относительно подстановки замощения. Подходит не любая подстановка. C1 а) Приведите пример такой подстановки, что все бесконечно декодируемые замощения непериодичны, но множество бесконечно декодируемых замощений не задаётся локальными правилами. b) Приведите пример такой подстановки, что все бесконечно декодируемые замощения периодичны. Искать подходящие подстановки проще не «с нуля», а в некотором смысле улучшая не подходящие. Переход к другому алфавиту. Декорации. Пусть задан алфавит плиток $a, b, c \ldots$ Мы можем ввести конечное число дубликатов (оттенков) для каждой буквы и рассмотреть расширенный алфавит плиток $a_{1}, \ldots a_{k}, b_{1}, \ldots b_{k}, c_{1}, \ldots c_{k}, \ldots$ Будем теперь вводить локальные правила для расширенного алфавита. После этого можно рассмотреть разрешенные замощения и игнорировать введенные оттенки. Такой прием называется введение декораций. Определение. Пусть задана подстановка $\sigma_{1}$ в алфавите $A_{2}$ и $\sigma_{2}$ в алфавите $A_{2}$. Подстановку $\sigma_{2}$ будем считать декорацией для $\sigma_{1}$, если выполнены следующие условия: 1) существует функция $f$, определенная на $A_{2}$, со значениями в $A_{1}$; 2) если $\sigma_{2}(a)=M$ - квадрат, составленный из букв алфавита $A_{2}$, то $\sigma_{1}(f(a))=f(M)$. Функцию $f$ можно понимать как забывание оттенка у цвета, а $f(M)$ - это квадрат того же размера, что и $M$, но из основных цветов без оттенков. Основной нашей задачей будет доказательство того, что заданную подстановку можно декорировать так, чтобы множество бесконечно декодируемых замощений задавалось локальными правилами. Для начала мы разберем некоторые полезные частные случаи. Пусть задана подстановка $\sigma$. Если $A-$ буква алфавита, то $\sigma(A)$ - квадрат $2 \times 2$ из четырех букв алфавита (куда отображается $A$ ). Будем считать, что разные символы отображаются в разные квадраты $2 \times 2$. Разделение образов. Определение. Определим отображение $\sigma_{U L}$, сопоставляющее символу $A$ левый верхний угол квадрата $\sigma(A)$. Аналогично определим отображения $\sigma_{D L}, \sigma_{U R}, \sigma_{D R}$. Будем считать, что подстановка обладает свойством разделения образов, если каждый символ алфавита присутствует в образе ровно одного из этих четырех отображений. C2 Приведите пример подстановки с разделением образов. Свойство разделения образов помогает нам при конструировании локальных правил. Сначала постараемся добиться цели для частного случая - какой нибудь подстановки. После этого, попробуем разобраться с подстановками с разделением образов. И уже потом перейдем к общему случаю. Идея построения. Для начала попробуйте раскрашивать ребра квадратов в различные цвета и формулировать локальные правила в терминах сочетаний этих цветов. Основная цель - добиться того, чтобы любое разрешенное замощение позволяло себя декодировать, то есть переходить к предыдущему уровню иерархии. Такой переход к предыдущему уровню и осуществляет подстановка. C3 Придумайте какую нибудь подстановку $\sigma$ и локальные правила, такие, что любое разрешенное замощение $S$ допускает переход к прообразу $\sigma^{-1}(S)$ и замощение $\sigma^{-1}(S)$ также является разрешенным. Докажите, что всякое разрешенное замощение является непериодичным. C4 Пусть задана $2 \times 2$ подстановка $\sigma$ с разделением образов. Пусть также заданы локальные правила, что любое разрешенное замощение $S$ допускает переход к прообразу $\sigma^{-1}(S)$ и замощение $\sigma^{-1}(S)$ также является разрешенным. Докажите, что всякое разрешенное данными локальными правилами замощение является непериодичным. C5 Пусть задана $2 \times 2$ подстановка $\sigma$ с разделением образов. Докажите, что для некоторой декорации подстановки множество бесконечно декодируемых замощений определяется локальными правилами. C6 Пусть задана $3 \times 3$ подстановка $\sigma$ с разделением образов. Докажите, что для некоторой декорации подстановки множество бесконечно декодируемых замощений определяется локальными правилами. C7 Пусть задана $2 \times 2$ подстановка $\sigma$, не обязательно с разделением образов. Докажите, что для некоторой декорации подстановки множество бесконечно декодируемых замощений определяется локальными правилами. ## D. РАЗЛИЧНЫЕ ФОРМАЛИЗМЫ И ПЕРЕВОД НЕПЕРИОДИЧНОСТИ НА ДРУГИЕ ЯЗЫКИ. Задачи о замощения можно формулировать на разных языках (формализмах). Иногда бывает полезно, получив продвижение в какой-то одной постановке, перевести ее на другую. Формализм Ванга. Каждая плитка - единичный квадрат. Каждая сторона квадрата раскрашена в один из конечного множества цветов. Квадраты можно прикладывать друг к другу, совмещая стороны одного цвета. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-143.jpg?height=150&width=742&top_left_y=586&top_left_x=739) Формализм дополняющих цветов. Каждая плитка - единичный квадрат. На сторонах каждого квадрата написаны ненулевые целые числа из конечного набора, которые можно назвать цветами. Квадраты можно прикладывать друг к другу, совмещая стороны, на которых написаны противоположные числа. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-143.jpg?height=146&width=742&top_left_y=870&top_left_x=738) Иногда считают, что вместе с плиткой в наборе есть и повернутая на 90 градусов плитка. В этом случае говорят, что плитки можно поворачивать. Рассмотрим плитку $A$ с цветами ( $\left.a_{1}, a_{2}, a_{3}, a_{4}\right)$ по часовой стрелке. Перевернутой плиткой будем называть плитку с цветами ( $-a_{3},-a_{2},-a_{1},-a_{4}$ ) против часовой стрелки. Если для каждой плитки в наборе есть и ее перевернутая плитка то говорят, что плитки можно переворачивать. Определение. Будем называть набор апериодическим (в каком-либо формализме), если с его помощью можно замостить плоскость только непериодически. D1 Докажите, что формализмы Ванга и дополняющих цветов эквивалентны, то есть если есть непериодический набор в одном формализме, то есть соответствующий ему непериодический набор в другом. D2 Назовем многоугольник квадратно-составленным, если он представляет собой фигуру полимино, составленную из единичных квадратов (связную и без дыр внутри). Постройте апериодический набор из квадратно-составленных многоугольников, которые можно поворачивать и переворачивать. D3 Рассмотрим формализм дополняющих цветов, причем плитки можно поворачивать и переворачивать. a) Может ли набор быть апериодичным? b) Пусть теперь по-прежнему плитки разрешается прикладывать по правилам дополняющих цветов, но перевернутые плитки не могут иметь общую сторону. Докажите, что существует апериодический набор. ## Замощения: подстановки и декорации. Решения. ## А. ОДнОмерный СЛУчАй A1 Omвет: 3. Понятно, что замощерия из всех одинаковых букв (единиц или нулей) подходят. Если в замощении есть и 1 , и 0 , то должны быть соседние 1 и 0 , причём 1 стоит слева. Далее понятно, что левее единицы могут стоять только единицы, а правее нуля - только нули. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-144.jpg?height=170&width=1328&top_left_y=430&top_left_x=344) A2 Примеров очень много. Например, если в одной клетке стоит 0 , а в остальных - единицы, то при любом сдвиге этот ноль перейдёт в единицу. | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | А3 Предлагаем убедиться, что если запретить слова $a a, a c, b a, b b, c b$ и $c c$, то у любой буквы однозначно определяются оба соседа и получается именно то, что нам нужно. A4 Omвет: $\partial a$. Например, можно ввести такие запреты, чтобы разрешёнными были периодические замощения с периодами $01,001, \ldots, 0^{100} 1$. Запретим слова $11,0^{101}$, а также $100 \cdot 99$ слов вида $10^{a} 10^{b} 1$ при всевозможных неравных $a$ и $b$ от 1 до 100 . Первые два запрета вынуждают, чтобы между двумя соседними единицами было от 1 до 100 нулей. Остальные запреты гарантируют, что слева и справа от любой единицы подряд идущих нулей поровну. A5 Напомним, что циклическим сдвигом слова $a_{1} a_{2} \ldots a_{n}$ называется любое из слов $a_{i+1} a_{i+2} \ldots a_{n} a_{1} a_{2} \ldots a_{i}$ при $0 \leqslant in$. Рассмотрим замощение $A$, состоящее из одних единиц. Оно не декодируемое, поэтому и замощение $s^{k}(A)$ не является бесконечно декодируемым. Покажем, что $s^{k}(A)$ допустимое. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-148.jpg?height=409&width=403&top_left_y=1449&top_left_x=384) Фрагмент замощения $s^{k}(A)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $$ s^{k+1}(1) $$ Видно, что любой квадрат со стороной не более чем $n$, встречающийся в $s^{k}(A)$, встречается в какомто блоке, составленном из четырёх квадратов $s^{k}(1)$. Но этот блок содержится внутри $s^{k+1}(1)$, а значит содержится внутри ковров Серпинского и не содержит запрещённых квадратов. В12 Пусть размер подстановки $\sigma$ равен $n$. Замощение $\sigma(A)$ состоит из квадратов $n \times n$ - образов от плиткобукв. Любой квадрат $2 \times 2$ находится в квадрате $2 n \times 2 n$ - объединении четырёх таких квадратов, то есть $\sigma\left(\begin{array}{ll}a_{1} & a_{2} \\ a_{3} & a_{4}\end{array}\right)$. Так как квадрат $\begin{array}{cc}a_{1} & a_{2} \\ a_{3} & a_{4}\end{array}$ разрешённый и $\sigma$ согласована с локальными правилами, то $\sigma\left(\begin{array}{ll}a_{1} & a_{2} \\ a_{3} & a_{4}\end{array}\right)$ не содержит запрещённых квадратов. B13 Рассмотрим какую-нибудь плитко-букву $a$ и последовательность квадратов $$ A_{0}=a, A_{1}=\sigma(a), A_{2}=\sigma^{2}(a) \ldots $$ В этой последовательность $A_{i}$ имеет размер $n^{i} \times n_{i}$, а из предыдущей задачи следует, что все $A_{i}-$ разрешённые. Из задачи $\mathbf{B 8}$ следует, что есть разрешённое замощение всей плоскости. B14 Oтвет: нет. Покажем, что в алфавите $\{0,1\}$ у подстановки $\sigma(0)=\frac{1}{1} \frac{1}{1}, \sigma(1)=\begin{array}{ll}0 & 0 \\ 0 & 0\end{array}$ любо замощение переходит в неэквивалентное ему. Предположим противное: $A$ - замощение, и $\sigma(A)$ эквивалентно $A$. Назовём отрезком длины $k$ прямоугольник $1 \times(k+2)$ или $(k+2) \times 1$, у которого две крайние клетки одного цвета, а остальные $k$ клеток - другого. Легко понять, что если в $A$ есть отрезки и минимальная длина отрезка равна $k_{0}$, то в $\sigma(A)$ минимальная длина отрезка - $2 k_{0}$. Значит, в $A$ отрезков нет вообще. Поэтому каждая линия (горизонталь или вертикаль) либо одноцветная, либо одна её половина заполнена единицами, а другая - нулями. Замощение $A$ не одноцветное, значит в нём есть соседние 1 и 0 . Не умаляя общности, 1 слева от 0 . Тогда в $\sigma(A)$ есть 0 , стоящий левее единицы. А так как в $A$ нет отрезков, то найдутся вершины прямоугольника, раскрашенные в шахматном порядке, например, так: ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-149.jpg?height=428&width=629&top_left_y=808&top_left_x=344) Далее получаем, что все клетки в том же ряду, что и $1^{A}$ и левее, заполнены единицами, а все клетки в том же ряду, что $0^{B}$ и левее, заполнены нулями. Тогда прямой угол с вершиной в клетке $1^{A}$, смотрящий вверх и налево, заполнен единицами. Но в таком случае в $\sigma(A)$ так же ориентированный угол заполнен нулями, чего не может быть. Критерий на существование замощения. (набросок) У подстановки $\sigma$ есть неподвижная точка тогда и только тогда, когда одна буква находится строго внутри своего образа ИЛИ две буквы находятся внутри своих образов на противоположных сторонах квадратов на одинаковом расстоянии от края ИЛИ четыре буквы появляются в углах своих образов во всех возможных позициях. Пусть $\sigma$ размера $k \times k$. Прежде всего заметим, что требование $S=\sigma(S)$ использовало бы задание системы координат, если же мы просто говорим, что $S$ эквивалентно $\sigma(S)$, то этого не нужно. Квадратик $1 \times 1$ переходит в квадрат $k \times k$ композицией гомотетии и сдвига. Это снова гомотетия, её неподвижная точка где-то находится. Она может быть в центре клетки, на границе двух клеток или в узле сетки, эти три случая соответствуют трём описанным ситуациям. С другой стороны, пусть $\sigma(a)$ содержит $a$ строго внутри себя, например в позиции $i, j$. Тогда $\sigma^{t}(a)$ можно рассматривать как квадратный паттерн размера $k^{t} \times k^{t}$, и мы можем указать их как $-k^{t} i /(k-$ $1) \cdots k^{t}(k-1-i) /(k-1)$ (аналогично по ординате) так, что $a$ это клетка с координатами 0,0 . Квадрат растёт во все стороны, и в итоге любая клетка $(x, y)$ покрыта каким-то квадратом. Каждый следующий квадрат содержит предыдущий, квадрат $\sigma^{t}(a)$ встречаестя в середине $\sigma^{t^{\prime}}(a)$ при $t^{\prime} \geq t$. поэтому можно определить замощение $S$ в клетке $(x, y)$ как букву $\sigma^{t}(a)(x, y)$ дял достаточно больших $t$. По построению $\sigma(S)$ будет сдвигом $S$. Другие случаи (две буквы на сторонах и 4 буквы в углу) разбираются аналогично. ## С. ПРИМЕРЫ НЕПЕРИОДИЧНОСТИ C1 а) Например, такая подстановка $\sigma$ : ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-149.jpg?height=173&width=821&top_left_y=2307&top_left_x=346) То, что множество бесконечно декодируемых замощений не задаётся локальными правилами, доказывается так же, как и для ковров Серпинского. Вот такую фигуру (а также если поменять местами 0 и 1) будем называть двойной рамкой. | | | | | | | | | | | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | | | 1 | 1 | 1 | 1 | 1 | 1 | 1 | | | | 1 | 0 | 0 | 0 | 0 | 0 | 1 | | | | 1 | 0 | | | | 0 | 1 | | | | 1 | 0 | | | | 0 | 1 | | | | 1 | 0 | | | | 0 | 1 | | | | 1 | 0 | 0 | 0 | 0 | 0 | 1 | | | | 1 | 1 | 1 | 1 | 1 | 1 | 1 | | | | | | | | | | | | Будем говорить, что размер двойной рамки на рисунке равен 5. Если в замощении встречается двойная рамка размера $N$, то у замощения не может быть вектора периодичности с абсолютными значениями координат меньшими $N$. Индукцией по $k$ показывается, что $\sigma^{k}(0)$ и $\sigma^{k}(1)$ содержат двойные рамки размера $3^{k-1}$. b) Пусть, например, в алфавите всего одна буква. C2 Пусть в плиточном алфавите 8 букв $a_{i}, b_{i}, c_{i}, d_{i}$, где индекс $i$ принимает значения 1 и 2 . Образы всех букв будут $\begin{aligned} & a_{i} b_{j} \\ & c_{k} d_{l}\end{aligned}$. Есть 16 способов выбрать индексы, можно взять разные способы для всех восьми букв. C3 Следует из следующих задач. C4 Докажем непериодичность бесконечно декодируемого замощения $A$. Прежде всего заметим, что из свойства разделения образов следует, что обе координаты вектора периодичности $v$ должны делиться на 2 , а потом заметим, что вектор $v / 2$ является вектором периодичности для замощения $\sigma^{-1}(A)$. Ненулевой целочисленный вектор можно делить пополам конечное число раз, после чего придём к противоречию. C5 Обозначения: влево $-W$, вправо $-E$, вверх $-N$, вниз $-S$. Влево-вверх будем обозначать $N W$, аналогично с другими диагональными направлениями. Считаем, что у подстановки со свойством разделения образов каждая буква $a_{i}$ плиточного алфавита $A$ относится к одному из типов $N W, N E, S W, S E$. Буквы нового алфавита $B$ (декорации) будем рисовать как плитки, в центре которых написаны буквы из $A$, а на сторонах проведены стрелки и на каждой стороне написано по метке - букве из $A$. Таким образом, буква из $B$ задаётся упорядоченной пятеркой букв из $A$ и четырьмя битами (направления стрелок). ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-150.jpg?height=122&width=119&top_left_y=1457&top_left_x=363) В дальнейшем будем говорить о типе плитки $(N E, N W, S E, S W)$, типе стороны плитки $(N, E, S, W$ в зависимости от направления стрелки), типе метки на стороне ( $N E, N W, S E, S W)$. Подстановка $\psi$ на алфавите $B$ определяется следующим образом. Если в центре плитки записана буква $a$ и какие-то метки по краям, и $\sigma(a)={ }_{a_{3}}^{a_{1}} a_{2}$, то в центре плиток образа записаны соответственные буквы алфавита, крайние рёбра образа $2 \times 2$ имеют такие же метки и направления, что и соответствующее ребро прообраза, а центральные четыре ребра исходят из центра квадрата $2 \times 2$ и на них метки $a$, смотри рисунок: ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-150.jpg?height=255&width=634&top_left_y=1897&top_left_x=347) Смысл такой декорации в следующем: бесконечно декодируемое замощение состоит из блоков $2^{k} \times 2^{k}$, а каждый блок $2^{k} \times 2^{k}$ - из четырёх блоков $2^{k-1} \times 2^{k-1}$, и границы между этими четырьмя блоками образуют крест, на рёбрах которого написана одна и та же буква. Эти кресты разграничивают блоки большого размера друг от друга. В примере с коврами Серпинского такой границы не было, и локальными правилами нельзя было отличить границу большого блока от его средней части. Локальные правила строились по такому принципу: изучалось, что в бесконечно декодируемом замощении может находиться на стыке двух и четырёх плиток, и эти свойства записывались в правила. Нужно, с одной стороны, записать достаточно правил, чтобы можно было определить декодирование, а с другой стороны - проверить, что после декодирования все записанные правила выполняются. По этой причине мы не используем больших локальных правил, про далёкие плитки: сложно проверять их выполнение после декодирования. Будем говорить, что направление $N E$ симметрично относительно вертикали направлению $N W$, а $S W$ симметрично относительно вертикали направлению $S E$. Симметрия относительно горизонтали определяется аналогично. Итак, локальные правила: 1) На соседних плитках смежные стороны имеют одинаковые направление и метку (то есть можно говорить о направлении и метке ребра во всём замощении). 2) Граничить по вертикальной стороне могут только плитки типов, симметричных относительно вертикали. Граничить по горизонтальной стороне могут только плитки типов, симметричных относительно горизонтали. 3) Узлы между типов могут иметь только такие степени вхождения: 4 исходят или 3 входит и одно исходит. 4) Если из узла 4 ребра исходят, то на них должны быть одинаковые метки. 5) Если слева-сверху от узла находится плитка типа $N W$, назовём такой узел центральным. Потребуем, чтобы из центрального узла выходило 4 ребра. Если на этих рёбрах метки $a$, и при этом $\sigma(a)={ }_{a_{1}}^{a_{1}} a_{2}$, то в центре плитки к северо-западу от узла должна быть буква $a_{1}$, аналогично про три другие плитки. 6) В узле типа 3-1 есть исходящее ребро, центральное входящее ребро, и два боковых входящих. Потребуем, чтобы метки на центральном входящем и на исходящем совпадали, а также чтобы типы меток на боковых входящих были симметричны относительно исходящего ребра. Например так: ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-151.jpg?height=209&width=208&top_left_y=936&top_left_x=384) 7) Пусть есть узел типа $3-1$, на исходящем ребре $v_{1}$ метка $a$, на боковом входящем ребре $v_{2}$ метка $b$ ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-151.jpg?height=43&width=244&top_left_y=1193&top_left_x=382) Если угол между типом $b$ и направлением $v_{2}$ равен $135^{\circ}$ (например, $v_{2}$ идёт вниз, а $b$ типа $N E$ ), то угол между $v_{1}$ и типом $b$ должен равняться $45^{\circ}$, а сама буква $b$ должна встречаться в $\sigma(a)$ на месте, соответствующем типу $b$. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-151.jpg?height=216&width=1003&top_left_y=1353&top_left_x=385) Анализ разрешённых замощений. Теперь нужно понять, какие бывают замощения с данными локальными правилами. Из 1) получаем, что плоскость разбивается на квадраты $2 \times 2$, в каждом из которых типы плиток ${ }_{S W}^{N W} \underset{S E}{N E}$. Далее будем их называть базовыми блоками. Из 5) и 4) следует, что центры базовых блоков - центральные узлы, из них всех выходит по 4 стрелки с одинаковыми буквами. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-151.jpg?height=384&width=393&top_left_y=1830&top_left_x=367) Все узлы разделим на центральные, боковые (те, в которые на картинке выше ведут по 2 стрелки) и остальные. Из 3) следует, что середины сторон базовых блоков (боковые узлы) имеют тип $3-1$, а вместе с 6) это даёт, что на каждой стороне базового блока стрелки сонаправлены и на них одинаковые метки. Вместе со свойством 4) это даёт, что каждый базовый блок является образом какой-то буквы при подстановке $\psi$, таким образом, однозначно определён переход от разрешённого замощения $T$ к $\psi^{-1}(T)$. Это замощение можно получить из $T$, стирая линии внутри базовых блоков и записывая внутрь блока ту букву, которая является меткой всех рёбер, выходящих из его центра. Нужно проверить для $\psi^{-1}(T)$ свойства 1) - -7). Свойства 1), 3), 4) и 6) выполняются автоматически, так как новых ситуаций в узлах (какие бетки и стрелки написаны на входящих рёбрах) не возникает. Для свойства 2) надо проверить, что любые два базовых блока, имеющие общую сторону, декодируются в буквы симметричных типов. Но тип после декодирования совпадает с типом метки, написанной на внутреннем кресте базового блока. Локальное правило 5) гарантирует, что метки у двух блоков симметричны, и после декодирования выполняется 2). А значит, базовые блоки объединяются в четвёрки, которые после декодирования имеют вид $S W S E$. Рассмотрим одну такую четвёрку. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-152.jpg?height=403&width=420&top_left_y=411&top_left_x=359) Метка $a$ имеет тип $N W$, поэтому, согласно 6 ), ребро $v_{1}$ направлено влево, а $v_{2}$ - вверх. Значит, $v_{1}$ и $v_{2}$ выходят из узла, из которого 4 выходящих ребра. Из 4) следует, что метки на этих рёбрах совпадают. Также из 6) следует, что в верхнем левом углу образа этой метки при $\sigma$ стоит $a$. Аналогично и про остальные базовые блоки, а значит, после декодирования выполняется свойство 5). Итак, декодированное замощение является разрешённым, а значит, все разрешённые замощения - бесконечно декодируемые. Вообще говоря, задачу мы пока решили не до конца. Мы не показали, что каждое бесконечно декодируемое при помощи $\sigma$ замощение допускает разрешённую декорацию. Можно показать, что такая декорация существует для тех замощений, у которых каждая плитка находится строго внутри некоторого блока вида $\sigma^{k}\left(a_{i}\right)$. Можно описать остальные замощения и слегка подправить нашу конструкцию, оставляем это читателю в качестве упражнения. C6 Эту задачу мы оставляем в качестве упражнения. Как тебе такое, Питер Шольц? C7 Воспользуемся задачей $C 5$. Пусть наша подстановка $\sigma$ над алфавитом $A$. Возьмём подстановку $\tau 2 \times 2$ с разделением образов над алфавитом $B$ (такая существует, см. задачу $C 2$ ). Рассмотрим алфавит $C$ размера $|A| \cdot|B|$, буквы которого - пары букв $\left(a_{i}, b_{j}\right) \mid a_{i} \in A, b_{j} \in B$. Определим на нём подстановку $\psi$ ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-152.jpg?height=63&width=1673&top_left_y=1432&top_left_x=302) она является декорацией $\sigma$ и у неё есть свойство разделения образов. ## D. РАЗЛИЧНЫЕ ФОРМАЛИЗМЫ И ПЕРЕВОД НЕПЕРИОДИЧНОСТИ НА ДРУГИЕ ЯЗЫКИ. D1 Допустим, есть апериодический набор в формализме Ванга. Введем для каждой буквы на ребре противоположную. Теперь в каждой плитке набора поменяем буквы верхнего и правого ребра на противоположные. Получится набор в другом формализме, причем каждому замощению в одном наборе соответствует замощение в другом. То есть апериодичность сохраняется. Аналогично в другую сторону. D2 Назовем многоугольник квадратно-составленным, если он представляет собой фигуру полимино, составленную из единичных квадратов (связную и без дыр внутри). Постройте апериодический набор из квадратно-составленных многоугольников, которые можно поворачивать и переворачивать. Указание. Такой набор можно построить, используя набор Робинсона. Сначала представим что каждая плитка это квадрат $k \times k$ для большого $k$. Реализуем каждую засечку на стороне квадрата из набора робинсона (их три вида) в виде выреза по контуру из D3 а) Oтвет: нет. b) Пусть теперь по-прежнему плитки разрешается прикладывать по правилам дополняющих цветов, но перевернутые плитки не могут иметь общую сторону. Докажите, что существует апериодический набор. ## Е. ДЕКОРАцИи E1. а) алфавит $\left\{a_{1}, a_{2}, b\right\}$; запретим $\left\{b a_{1}, a_{2} b, b b, a_{2} a_{1}\right\}$. b) Алфавит $\left\{a_{1}, a_{2}, b\right\}$; запретим горизонтальные пары $\left\{b a_{1}, a_{2} b, b b, a_{2} a_{1}\right\}$, вертикальные пары $\left\{\begin{array}{c}b, a_{1} b a_{2} \\ a_{2}{ }^{\prime} b^{\prime}, b a_{1}\end{array}\right\}$, а также $\left\{\begin{array}{ll}a_{1} & a_{2} \\ a_{2} & \end{array}\right.$. E2. Плитками без $b$ (декорированными) можно замостить сколь угодно большой квадрат. Согласно задаче B8, можно замостить всю плоскость, не используя $b$. E3. a) Oтвет: нет. Допустим противное, пусть $k$ - размер максимального запрета. Рассмотрим замощение, в котором одна компонента связности - горизонтальный путь длины $k$, и другое замощение - путь длины $k+1$. У этих замощений одинаковый набор паттернов размера $k \times k$, поэтому они оба должны быть или разрешены, или запрещены. А должно быть только одно из них. b) алфавит $\left\{a, b_{1}, b_{2}\right\}$; запреты $\left\{a b_{2}, b_{1} b_{1}, b_{2} b_{2}, b_{1} a\right\}$. c) Набросок одного из возможных решений. Пусть декорации изображают остовный лес для связной компоненты, то есть объединение непересекающихся деревьев, у каждого дерева отмечен корень и каждая вершина компоненты входит ровно в одно дерево. Наложим условие на чётность числа вершин в компоненте: каждая вершина может считать по модулю 2 число вершин у своих сыновей, прибавлять единицу и передавать результат родителю (считаем, что дальше от корня находятся сыновья, а ближе к корню - родитель). С одной стороны, в каждой связной компоненте есть остовное дерево с чётным числом вершин, которое можно так раскрасить. С другой - в каждом разрешённом замощении каждая связная компонента состоит из одного или нескольких таких деревьев, т.е. содержит чётное число вершин. E4*. В размерности 1 это просто: алфавит $\left\{a, b_{1}, b_{2}\right\}$; запреты $\left\{a b_{2}, b_{1} b_{1}, b_{2} b_{2}, b_{2} a\right\}$. В размерности 2 это тоже возможно, но весьма и весьма нетривиально :-) E5. Пример: возьмём такой набор плиток, чтобы с их помощью можно было нарисовать на плоскости любой граф. Тогда декорацией можно задать правильную раскрашиваемость графа в 3 цвета. E6. а) Все квадраты $2 \times 2$ появляются или и там и там, или нигде. b) Пусть $k$ - размер декорированного алфавита. Декораций границы квадрата $n \times n$ не более чем $k^{4(n-1)}$. c) Рассмотрим множество замощений, где $c$-шки образуют колонку. Смотрим множество паттернов $n \times n$, которые могут быть слева от колонки. Их хотя бы $2^{n^{2}}$ (декорации всевозможных замощений из $a$ и $b$ ). При достаточно большом $n$ ( параметр $k$ фиксирован) есть два таких замощения, что у них совпадают границы, и по пункту а) одно можно заменить на второе. Тогда мы получим разрешённое, но не симметричное. d) Если $\mathcal{S}$ - такое множество замощений, то должно существовать такое $k$, что для любого $n$ среди любого набора из $k^{n}$ замощений квадрата $n \times n$ можно выбрать 2 и одинаково продолжить до замощения всей плоскости. # Tilings: substitutions and decorations Alexei Belov, Pierre Guillon, Ilya Ivanov-Pogodaev, Ivan Mitrofanov This project is devoted to plane tilings. Usually, if we can tile a plane with some types of tiles, then we use construction which structure is periodic. Correctly, a tiling is called periodical if it doesn't change after shift on non-zero distance. In 1961 chinese mathematician Hao Wang have stated the following problem: Consider unit squares. Each square's side is colored by one color of several possible ones. Consider some types of these squares. We can attach any square to another side to side if these sides have the same color. Suppose that we can use unlimited number of each type's squares. Is it true that if we can tile a plane then we can do it periodically? Firstly Wang expected that the answer is positive, namely if there exists some tiling then there is a periodic one. But in 1966 Robert Berger (student of Wang) have constructed a set consisting of 20426 squares such that there exist non periodical tilings only. Lately Berger reduced his set to 104 squares, and in 1971 Raphael Robinson greatly simplified the construction and introduced a set of six poligons such that one can tile a plane using them only non periodically. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-154.jpg?height=379&width=550&top_left_y=786&top_left_x=836) FiguRE 1. Robinson's tiles. This problem has some philosophical background. The point is to reach global condition (non periodicity) using local sources (boundary conditions). There are many such problems. How to organize a computation network such that any local fault can not violate global processes. Or how interaction of molecules leads to crystals growing. In this project we will study some ideas and methods to solve such problems in various settings. The main problem of first half is to construct a set of polygons such that there exist non periodical tilings only (problem B10). In fact, this problem can be reduced to constuction of local rules: boundary conditions which define when we can attach tiles to each other. This construction uses some ideas. In B and C we study some tricks that can be used to construct aperiodic tilesets. ## A. One dimensional case Consider two-side infinite tape which consists of cells (equal squares). We will assign one letter of the finite alphabet for each cell. A cell with assigned letter is called a tile. If every cell contains some letter (or the tape is covered by non intersecting tiles) then we say that this is the tiling. A tiling is called periodical if it doesn't change after shift on non zero integer $p$. It is easy to see that periodical tiling is actually periodical repeat of some combination of $p$ tiles. This combination is called period of tiling, $p$ is length of period. A forbidden word or local rule is a combination of several adjacent tiles. We will consider tilings which contain no forbidden words. In notations we use letters of a finite alphabet to denote tiles and we do words to denote combinations of tiles. Tilings which contain no forbidden words are called permitted ones. Further we assume that the number of local rules is finite. Two tiling which transforms to each other after some shift are called equivalent. Below we consider such tilings as the same. A1 Consider an alphabet $\{0,1\}$. The word 01 is forbidden. How many non-equivalent tilings exist? A2 Show that for two letters there exists a non periodical tiling. A3 Construct a finite set of forbidden words in the alphabet $\{a, b, c\}$, such that the periodic tiling with period $a b c$ is the only permitted tiling. A4 Does there exist a set of local rules such that there are exactly 100 nonequivalent permitted tilings? A5 Let $A$ be a finite word. Prove that there exists a of forbidden words such that the tiling with period $A$ is the only permitted one. A6 Let $A$ be an aperiodic tiling. Prove that there is no a finite set of forbidden words such that $A$ is permitted and all the permitted tilings are equivalent to $A$. A7 Consider an alphabet with size $n$. Suppose that there are $k$ forbidden words, and each of them has length 2 . Also, there are no permitted tilings. Find the minimal possible $k$. ## B. Dimension two: lOcal Rules and substitutions Conclusion from one dimensional case. In one dimensional case forbidden word is some kind of local condition. Using these conditions we want to obtain some global property. But in one dimensional case finite number of local rules does not allow us to force non periodical tilings. Let us proceed to two dimensional case. We will start with basic definitions and notation. Consider an infinite square grid and a finite alphabet $L$ (alphabet of tiles). We will write letters of $L$ into cells of an infinite square grid, one letter to one cell. This alignment is called a tiling. Similarly to one dimensional case, a tiling is called periodic, if it doesn't change after translation by a non zero vector $(a, b)$, and is called aperiodic in the opposite case. Equivalence of tilings. Definition. Consider two tilings $S_{1}$ and $S_{2}$. Suppose that we can translate $S_{1}$ by integer vector and obtain $S_{2}$. In this case we say that $S_{1}$ and $S_{2}$ are equivalent. Remark. Different tilings of finite regions are never equivalent. Local rules. Definition. Let $n>1$ be an integer. Consider a set of all squares $n \times n$, compiled from letter-tiles. Formally, this set is $L^{n^{2}}$ (set of all $\left(n^{2}\right)$-tuples of letters from $L$ ). Let $R$ be a subset of this set. So, $R$ is a set of several squares $2 \times 2$ compiled by $L$ letters. Let us call these squares as forbidden ones. A tiling is called permitted by $R$ if there are no forbidden squares in the tiling. The squares from $R$ are also called local rules, the number $n$ is the size of the local rules. Using local rules we can force some properties of permitted tilings. Let's do some practice. B1 Suppose that the alphabet contains only two letters (black and white squares). Construct local rules such that the chessboard tiling is the only permitted tiling. B2 Suppose that the alphabet contains two letters and eight squares $2 \times 2$ are forbidden (look at the picture). For given $m$ and $n$, how many permitted tilings of a $m \times n$ rectangle exist? | 1 | 1 | | :--- | :--- | | 0 | 1 | | 1 | 1 | | :--- | :--- | | 1 | 0 | | 1 | 0 | | :--- | :--- | :--- | | 1 | 1 | | 0 | 1 | | :--- | :--- | :--- | | 1 | 1 | | 1 | 0 | | :--- | :--- | :--- | | 0 | 0 | | 0 | 1 | | 0 | 0 | | 0 | 0 | | 0 | 1 | | 0 | 0 | | :--- | :--- | | 1 | 0 | B3 Suppose that the alphabet contains two letters. The set of local rules is the same as in the previous problem, and also the square with 1 on main diagonal (see picture below) is forbidden. Classify all different permitted tilings of the infinite plane. | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 1 | B4 Suppose that alphabet contains four letters and a $2 \times 2$ square is permitted if and only if it consists of four different letters. How many permitted tilings do exist for the rectangle $m \times n$ ? B5 Consider a binary alphabet. Suppose that there are $k$ forbidden squares $2 \times 2$ and suppose that there are no permitted tilings. Find the minimal possible $k$. B6 Consider some set of local rules in the alphabet $\{0,1\}$. Suppose that the tiling on picture below is permitted (other cells except these nine contain 0). Prove that there exists infinite number of nonequivalent permitted tilings. | 0 | 1 | 0 | | :--- | :--- | :--- | | 1 | 0 | 1 | | 1 | 1 | 1 | This problem show that sometimes ${ }^{1}$ it is not possible to determine one precise tiling by local rules. But if we cannot force one tiling, maybe we can determine some set of similar tilings? We say that some set of tilings is defined by local rules if tilings from this set are the only permitted tilings according with these local rules. B7 A frame is a figure consisting of $2(m+n)-4$ boundary cells of a rectangle $m \times n$ for some $m, n>1$. We call a $\{0 ; 1\}$-tiling pretty, if union of some (maybe infinite) set of frames is filled with units, and zeroes are placed in the remaining cells. Suppose we have a finite set of local rules such that all pretty tilings are permitted. Prove that there are infinitely many non-equivalent permitted not pretty tilings. B8 Suppose we have a set of local rules, and suppose that for any positive $r$ we can tile an area including circle of radius $r$ such that there are no forbidden squares. Prove that we can tile the whole plane with the same condition.[^6] Using local rules we can obtain really complicated tilings. For example, we can force all the permitted tilings to be aperiodic. It is hard to solve this problem now and we recommend to return to it in $C$-part and use some additional methods. B9* Construct a set of local rules such that all permitted tilings are non periodic. B10* Construct a finite set of polygons such that all tilings are non-periodic. All polygons can be rotated and reflected; they should be arranged without holes or overlaps. Definition. Suppose that for any letter in the alphabet $L$ corresponds a some square $k \times k$ compiled by letters of this alphabet. This correspondence is called a substitution. For a given substitution $\sigma$ and a tiling $A$ of some (finite or infinite) region we can construct a new tiling $\sigma(A)$ by simultaneous replacement of tile-letters by corresponding squares $k \times k$. Starting with one tile and iterating this procedure, we obtain tilings of bigger and bigger squares. Definition. Consider a $k \times k$ substitution $\sigma$. Suppose that a tiling $S$ can be uniquely divided by horizontal and vertical lines to squares $k \times k$ such that each square is image of some letter (any square is $\sigma(a)$ for some letter $a)$. In this case we can write a letter instead of each $k \times k$ square ( $a$ instead of $\sigma(a)$ ). Thus we obtain a new tiling $\sigma^{-1}(S)$. If $\sigma^{-1}(S)$ is defined in a unique way, we say that for tiling $S$ we can find an inverse image $\sigma^{-1}(S)$. A tiling is called infinitely decodable by $\sigma$ if one can find inverse image any number of times. $$ 1 \rightarrow \begin{array}{|l|l|l|} \hline 1 & 1 & 1 \\ \hline 1 & 0 & 1 \\ \hline 1 & 1 & 1 \\ \hline \end{array} \quad 0 \quad \begin{array}{|l|l|l|} \hline 0 & 0 & 0 \\ \hline 0 & 0 & 0 \\ \hline 0 & 0 & 0 \\ \hline \end{array} $$ B11 Consider the substitution above. An infinitely decodable tiling for this substitution is called a Sierpinski carpet. a) Prove that there are infinitely many nonequivalent Sierpinski carpets. b) Prove that all Sierpinski carpets besides one of them are aperiodic tilings. c) Is the set of all Sierpinski carpets defined by local rules? If one wants to define infinitely decodable tilings with local rules, it is natural to require that a substitution map permitted tilings only to permitted tilings. Definition. We say that a substitution $\sigma$ agrees with local rules $R$ of size $k$ if the following conditions hold: 1) the square $\sigma(a)$ is permitted for every letter $a$; 2) if $X$ is permitted square then the $2 n \times 2 n$ square $\sigma(X)$ contains only permitted $k \times k$ squares. B12 Consider some local rules $R$. Suppose that $k=2$ (all forbidden squares are $2 \times 2$ ). Also, suppose that a $N \times N$ square $A$ is permitted (it contains no forbidden $2 \times 2$ squares). Let a substitution $\sigma$ agree with local rules. Prove that the square $\sigma(A)$ is also permitted. B13 Suppose that substitution $\sigma$ agrees with the local rules of size 2. Prove that there exists a permitted tiling. B14 Consider a substitution $\sigma$. Does there exist a tiling $S$ such that $\sigma(S)$ is equivalent to $S$ ? Find sufficient conditions on $\sigma$ for existence of such a tiling. ## C. EXAMPLES OF NON PERIODICITY Definition. A tileset is aperiodic, if it admits at least one tiling of the plane but doesn't admit a periodic one. Conclusions from parts A and B. We want to find a set of tilings such that it is defined by local rules and contains only aperiodical tilings. There is no such an example in one-dimensional case (why?). The simplest two-dimensional examples we know are infinitely decodable tilings for some substitutions. Not any substitution is suitable. C1 a) Find a substitution such that all infinitely decodable tilings are aperiodical, but the set of them is not defined by local rules. b) Find a substitution such that all infinitely decodable tilings are periodical. It is not easy to find a suitable substitution "from scratch", so we'll "improve" not suitable ones. Transition to another alphabet. Decorations. Consider a tile alphabet $a, b, c \ldots$ We can make finite number of duplicates (shades) for each letter and consider the extended alphabet of tiles $a_{1}, \ldots a_{k}, b_{1}, \ldots b_{k}, c_{1}, \ldots c_{k}, \ldots$ Now we can set up local rules for this extended alphabet. After that we can take permitted tilings and ignore the shades. This approach is called setting up decorations. Definition. Let $A_{1}$ and $A_{2}$ be two tile alphabets and let $\sigma_{1}$ and $\sigma_{2}$ be two substitutions in alphabets $A_{1}$ and $A_{2}$ respectively. We call the substitution $\sigma_{2}$ a decoration for $\sigma_{1}$ if the following conditions are satisfied: 1) there exists a mapping $f$ from $A_{2}$ to $A_{1}$; 2) if $\sigma_{2}(a)=M$ is a square made of letters from $A_{2}$, then $\sigma_{1}(f(a))=f(M)$. We consider $f$ as "forgetting the shade", and $f(M)$ is a square obtained from $M$ by applying $f$ to each letter. Our main goal is to prove that a given substitution can be decorated in such a way that the set of infinitely decodable tilings is defined by local rules. Firstly we deal with some useful special cases. Consider a $2 \times 2$ substitution $\sigma$. If $a$ is a letter, then $\sigma(a)$ is a $2 \times 2$ square which consists of four letters. We assume that different letters correspond to different $2 \times 2$ squares. Images separating. Definition. Let us define $\sigma_{U L}$ mapping. It maps letter $a$ to the upper left corner of the square $\sigma(a)$. Similarly we can define mappings $\sigma_{D L}, \sigma_{U R}, \sigma_{D R}$. We say that the substitution separate the images (or has property of separating images) if every letter occurs exactly at one image of these four mappings. C2 Construct a substitution with property of separating images. The property of separating images helps us to construct local rules. Firstly we try to archive the goal for particular case - some fixed substitution. After this, we will study substitutions with separating images. And finally we will proceed to the common case. The idea of construction. Firstly try to color sides of squares in different colors and formulate local rules in terms of these colors combinations. The main goal is to obtain the property that any permitted tiling can be decoded that is to proceed to next level of hierarchy. A substitution make this transition. C3 Find some substitution $\sigma$ and local rules such that any permitted tiling $S$ can be decoded and tiling $\sigma^{-1}(S)$ is also permitted. Prove that any permitted tiling is aperiodical. C4 Consider a $2 \times 2$ substitution $\sigma$ with property of separating images. Suppose that there exist local rules $R$ such that for any permitted tiling $S$ there exists inverse image $\sigma^{-1}(S)$ which is also permitted by the same local rules $R$. Prove that any permitted tiling is aperiodic. C5 Consider a $2 \times 2$ substitution with property of separating images. Prove that there is a decoration such that the set of infinitely decodable tilings is defined by local rules. C6 Consider a $3 \times 3$ substitution with property of separating images. Prove that there is a decoration such that the set of infinitely decodable tilings is defined by local rules. C7 Consider a $2 \times 2$ substitution (maybe there is no separating images property). Prove that there is a decoration such that the set of infinitely decodable tilings is defined by local rules. ## D. OTHER FORMALISMS AND TRANSITIONS OF NON-PERIODICITY TO OTHER LANGUAGES In addition to local rules and forbidden squares, there are other formalisms to specify tilings. Sometimes the interaction of different formalisms is useful. Wang tiles formalism. There are finitely many colors, a Wang tile is a unit square with a color on each side. There is a finite set of Wang tiles, colors of common sides should match in tilings of the plane. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-157.jpg?height=128&width=670&top_left_y=1514&top_left_x=778) Complementary colors formalism. There is a finite set of non-zero integers (we call them colors). A tile is a unit square with an integer on each side. There is a finite set of tiles, numbers on common sides should be opposite. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-157.jpg?height=140&width=676&top_left_y=1778&top_left_x=774) D1 Prove that Wang's formalisms and complementary colors formalism are equivalent, i.e. if there is an aperiodic tileset in one formalism, then there is a corresponding aperiodic tileset in the other. Sometimes a tileset along with any tile contains the rotated on 90 degree tile. In this case we say that we can rotate tiles. Consider a tile $A$ with integers $\left(a_{1}, a_{2}, a_{3}, a_{4}\right)$ written on its sides clockwise. The tile with $\left(-a_{3},-a_{2},-a_{1},-a_{4}\right)$ integers written on the same sides is called flipped tile (with respect to $A$ ). If a tileset contains the flipped tile for every tile then we say that we can flip tiles. D2 We call a polygon square-composed, if it is connected polymino composed of unit squares without holes inside. Construct an aperiodic set of square-composed polygons such that with every polygon $\Phi$ the tileset contains all 8 rotations and reflections of $\Phi$. D3 Suppose that a tileset in complementary colors formalism with any tiles contains all its rotated and flipped tiles. a) Can such a tileset be aperiodic? b) Add extra rule to the formalism: a tile can not be a side-neighbor of its flipped tile. Prove that such a tileset can be aperiodic. ## Tilings: substitutions and decorations. Solutions. ## A. OnE Dimensional CaSE A1 Answer: 3. It is clear that two tilings with all identical letters (ones or zeros) are permitted. If there are both 1 and 0 , then there must be adjacent 1 and 0 , and 1 is on the left. It's clear that only 1 's can be to the left of 1 , and only 0 's can be to the right of 0 . ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-158.jpg?height=74&width=1322&top_left_y=456&top_left_x=346) | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | A2 There are a lot of examples. For instance, one 0, and the other cells are filled with 1-s. Then for any nonzero shift this zero will map to a different one. | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | A3 See for yourself that if you forbid the words $a a, a c, b a, b b, c b$, and $c c$, then any letter uniquely forces both of its neighbours. A4 Answer: yes. For example, we can forbid all tilings except 100 periodic sequences with periods $01,001, \ldots, 0^{100} 1$. We forbid $11,0^{101}$, and $100 \cdot 99$ words of form $10^{a} 10^{b} 1$ for all $a \neq b$ from 1 to 100. The first two rules force that the number of zeroes between two neighbouring 1's is from 1 to 100. The rest of them guarantee that any two neighbouring blocks of zeroes are equal. A5 Recall that a cyclic shift of a word $a_{1} a_{2} \ldots a_{n}$ is any one of the words $a_{i+1} a_{i+2} \ldots a_{n} a_{1} a_{2} \ldots a_{i}$ for $0 \leqslant in$. Consider a tiling $A$ of the plane by 1 's. It is not decodable, so the tiling $s^{k}(A)$ is not infinitely decodable. We show that the tiling $s^{k}(A)$ is permitted. ![](https://cdn.mathpix.com/cropped/2024_04_28_ff38bcac6327ce3425d3g-162.jpg?height=412&width=407&top_left_y=781&top_left_x=363) a pattern of $s^{k}(A)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- | | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(0)$ | $s^{k}(1)$ | | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $s^{k}(1)$ | $$ s^{k+1}(1) $$ It can be shown that any square with side no more than $n$ that occurs in $s^{k}(A)$, occurs in some block composed of four squares $s^{k}(1)$. But this block is located inside $s^{k+1}(1)$, and therefore it is a pattern of Sierpinski carpets and does not contain forbidden squares. B12 Let $n$ be the size of $\sigma$. The tiling $\sigma(A)$ is composed of $n \times n$-squares - images of letters. Any $2 \times 2$-square is located inside some $2 n \times 2 n$-square (union of 4 such squares), which is of the form $\sigma\left(\begin{array}{ll}a_{1} & a_{2} \\ a_{3} & a_{4}\end{array}\right)$. Since the square $\begin{array}{ll}a_{1} & a_{2} \\ a_{3} & a_{4}\end{array}$ is permitted and $\sigma$ agrees with local rules, $\sigma\left(\begin{array}{ll}a_{1} & a_{2} \\ a_{3} & a_{4}\end{array}\right)$ does not contain forbidden patterns. B13 Consider a tile with letter $a$ and a sequence of squares $$ A_{0}=a, A_{1}=\sigma(a), A_{2}=\sigma^{2}(a) \ldots $$ In this sequence $A_{i}$ has size $n^{i} \times n^{i}$, and it follows from the previous problem that $A_{i}$ is permitted for any $i$. From B8, it follows that there exists a permitted tiling of the whole plane. B14 Answer: not always. For alphabet $\{0,1\}$ and substitution $\sigma(0)=\frac{1}{1} \frac{1}{1}, \sigma(1)={ }_{0}^{0} 0$ such a tiling does not exist. The criterion for the existence of a fixed tiling. $\sigma$ admits a fixed tiling if and only if one letter appears strictly inside its own image, or two letters appear in their own images in some opposite side of the square, strictly, or four letters appear in their own images in all possible corners. Sketch of proof. Let $\sigma$ be a $k \times k$-substitution. First note that requiring $S=\sigma(S)$ would involve the choice of an origin tile, which is not a problem here: we just want $S$ and $\sigma(S)$ to be equivalent. This means that there exist integers $i, j$ such that, in $S$, the image by $\sigma$ of any cell $(x, y)$ is the pattern appearing at positions $(i+k x \cdots i+k x+k-1, j+k y \cdots j+k y+k-1)$. We can do a euclidean division of $i$ by $1-k$, and get $i=(1-k) x+i^{\prime}$ for some $i^{\prime}$ with $0 \leq i^{\prime}А. Еннэ, А. Скопенков, А. Чиликов * ## 1 Введение и формулировки результатов ## 1.1 О чём этот цикл задач Знаменитые теоремы ${ }^{1}$ Руффини 2.7 , Абеля и Галуа 1.3, 1.4 о неразрешимости алгебраических уравнений в радикалах - классический результат алгебры, интересный для информатики (теории символьных вычислений). Формулировки этих теорем приведены ниже. Основное содержание данного текста - изложение глубоких идей алгебры (точнее, теории Галуа) на красивых простых доказательствах этих теорем, см. [ZSS, §27]. Замечательно, что при этом для понимания приводимых доказательств достаточно уметь делить многочлены с остатком, извлекать корни из комплексных чисел, умножать перестановки и решать системы линейных уравнений. И тот, кто не дойдёт до полного доказательства основных резуль-[^7]татов, сможет порешать задачи для исследования, см. [E2, Es, AB, Ko17, Saf] и ссылки в этих работах. Перед доказательствами неразрешимости алгебраических уравнений мы разберем общий способ их решения - метод резольвент Лагранжа. Идея Абеля и Галуа фактически заключается в том, что если уравнение разрешимо в радикалах, то его можно решить этим методом. Этим же методом строятся и алгоритмы - например, распознаваемости разрешимости уравнений в радикалах. Для практики приближённые методы решения уравнений более полезны, чем радикальные формулы. Кроме того, уравнения можно решать при помощи трансцендентных функций (см. метод Виета [ZSS, п. 4.2] и [PSo]). Однако проблема разрешимости в радикалах интересна как пробная задача современных теорий символьных вычислений и сложности вычислений. О новизне. Приводимые в решениях доказательства не претендуют на новизну (хотя, возможно, читатели сумеют придумать что-то новое). Все же в этом тексте имеется много методических находок, см. [ZSS, п. 5.2.1, 5.2.2], и доказательства отличны от приведенных и цитированных в [ZSS, §5]. Однако, к сожалению, приводимые доказательства малоизвестны. Как следствие, малоизвестно, что не только решать квадратные и кубические уравнения, но и доказывать указанные теоремы экономнее, не строя и затем применяя теорию Галуа (как, например, в стандартных учебниках по алгебре), а напрямую ${ }^{2}$ - но при этом, конечно, переоткрывая и используя базовые идеи этой теории. ${ }^{2}$ Как, например, в [Dor, §25], [Pr07-2, дополнение 8], [FT, Лекция 5], [ZSS, §5], [Dor, St94, Kol, Ler, T, Sk11, Sk15] и здесь. Изложение в [Al] ближе к этому стилю. Хотя большая часть [Al] посвящена изложению теории, не нужной для доказательства ослабленной версии теоремы Абеля, объявленной в качестве основного результата (см. [Sk15, конец замечания 7]), автору книги [Al] удалось избежать немотивированного изложения части этой теории. Доказательство из [Al] более коротко и понятно изложено в [FT, Лекция 5] и, возможно, в [Sk11]. Заметим, что доказательства в большинстве этих источников неполны, см. [ZSS, сноска 12 на стр. 113 и конец §5.5.4], [Sk15, Обсуждение]. Несмотря на эти недостатки, вышеупомянутые элементарные изложения были для нас полезнее, чем формальные изложения (в стандартных учебниках, излагающих теории), которые начинаются с нескольких сотен страниц определений и следствий, роль которых в доказательстве теоремы о неразрешимости неясна на момент их формулировки. Немотивированное изложение служит «главным образом для ## 1.2 Неразрешимость в вещественных радикалах Вещественное число называется вещественно радикальным, если его можно получить из числа 1 при помощи сложений, вычитаний, умножений, делений на ненулевые числа и извлечений корней целых положительных степеней из положительных чисел. Т.е. если некоторое множество, его содержащее, можно получить из множества $\{1\}$, используя операции добавления к уже имеющемуся множеству $M \subset \mathbb{R}$, содержащему числа $x, y$, $$ \begin{gathered} \text { чисел } \quad x+y, x-y, x y, \quad \text { числа } x / y \text { при } y \neq 0 \\ \text { и числа } \sqrt[n]{x} \text { при } x>0 \text { и целом } n>0 . \end{gathered} $$ Вещественная радикальность числа $\alpha$ равносильна существованию таких - целых положительных чисел $s, k_{1}, \ldots, k_{s}$, - вещественных чисел $f_{1}, \ldots, f_{s}$ и многочленов $p_{0}, p_{1}, \ldots, p_{s}$ от $0,1, \ldots, s$ переменных, соответственно, с рациональными коэффициентами, что $$ \left\{\begin{array}{l} f_{1}^{k_{1}}=p_{0} \\ f_{2}^{k_{2}}=p_{1}\left(f_{1}\right) \\ \cdots \\ f_{s}^{k_{s}}=p_{s-1}\left(f_{1}, \ldots, f_{s-1}\right) \\ \alpha=p_{s}\left(f_{1}, \ldots, f_{s}\right) \end{array}\right. $$ Замечание 1.1. (а) Любой вещественный корень квадратного уравнения с рациональными коэффициентами вещественно радикален. (b) Уравнение $x^{3}+x+1=0$ имеет ровно один вещественный корень, который вещественно радикален [ZSS, п. 4.2], см. также задачу 2.8 (c). (c) Уравнение $x^{4}+4 x-1=0$ имеет два вещественных корня, каждый из которых вещественно радикален [ZSS, п. 4.2], см. также задачу $2.10(\mathrm{~d})$. того, чтобы затруднить непосвященным овладение своей наукой и тем самым повысить ее авторитет» [Ar84, стр. 49]. Заметим, что для многих именно мотивированное изложение повышает авторитет математики. (d) Любое вещественно построимое число [ZSS, п. 5.1.2] вещественно радикально. (е) Существует многочлен 3 -й степени с рациональными коэффициентами (например, $x^{3}-3 x+1$ ), ни один из корней которого не является вещественно радикальным. (Это доказано в п. (f).) (f) Число $\cos (2 \pi / 9)$ не является вещественно радикальным. Действительно, по формуле косинуса тройного угла каждое из чисел $\cos (2 \pi / 9), \cos (8 \pi / 9), \cos (14 \pi / 9)$ удовлетворяет уравнению $8 y^{3}-6 y+1=0$. По нижеприведенной теореме 1.2 ни одно из них не является вещественно радикальным. (g) Трисекция угла невозможна при помощи вещественных радикалов, т.е. существует такое $\alpha$ (например, $\alpha=2 \pi / 3$ ), что число $\cos \alpha$ вещественно радикально, а число $\cos (\alpha / 3)$ - нет. (Это следует из п. (f).) Теорема 1.2 (о разрешимости в вещественных радикалах). Следующие условия на многочлен $f$ третьей степени с рациональными коэффициентами равносильны: (i) многочлен $f$ имеет либо хотя бы один рациональный корень, либо ровно один вещественный корень; (ii) многочлен $f$ имеет вещественно радикальный корень; (iii) все вещественные корни многочлена $f$ вещественно радикальны. Единственность вещественного корня «укороченного» уравнения $x^{3}+p x+q=0$ равносильна условию $\ll p=q=0$ или $(p / 3)^{3}+$ $(q / 2)^{2}>0 »[\mathrm{ZSS}$, задача 8.1.5.d]. Равносильность $(i i) \Leftrightarrow(i i i)$ очевидна и следует из замечания 1.1.а. Разрешимость в теореме 1.2 (т.е. $(i) \Rightarrow(i i)$ ) доказывается методом дель Ферро [ZSS, п. 4.2]; см. другое доказательство в п. 2.3. Неразрешимость в теореме 1.2 (т.е. $(i i) \Rightarrow(i)$ ) доказывается сложнее. Более просто доказывается аналогичный результат о неразрешимости в многочленах, см. п. 2.5. ## 1.3 Неразрешимость в комплексных радикалах Перейдём к формулам, которые могут содержать комплексные числа. Оказывается, кубическое уравнение (например, $x^{3}-3 x+1$ ), не- разрешимое в вещественных радикалах, разрешимо в комплексных. Комплексное число называется (комплексно) радикальным, если его можно получить из числа 1 при помощи сложений, вычитаний, умножений, делений на ненулевые числа и извлечений корней целых положительных степеней. Т.е. если некоторое множество, его содержащее, можно получить из множества $\{1\}$, используя операции добавления к уже имеющемуся множеству $M$, содержащему числа $x, y$, $$ \text { чисел } x+y, x-y, x y, \quad \text { числа } x / y \text { при } y \neq 0 $$ и любого такого числа $r \in \mathbb{C}$, что $r^{n}=x$ для некоторого целого $n>0$. Например, любой (комплексный) корень квадратного уравнения с рациональными коэффициентами является радикальным. Аналогичные утверждения справедливы для уравнений 3 -й и 4 -й степени. Они доказываются методами делъ Ферро и Феррари [ZSS, п. 4.2]; см. другое доказательство в п. 2.3. Однако аналог этих утверждений для более высоких степеней неверен. Теорема 1.3 (Галуа). Существует уравнение 5-й степени с рациональными коэффициентами (например, $x^{5}-4 x+2=0$ ), ни один из корней которого не является радикальным. Знаменитую проблему о разрешимости уравнений в радикалах решили доказанные немного ранее более слабые теоремы РуффиниАбеля. Теорема Руффини 2.7 сложнее формулируется, но подводит нас к доказательству теоремы Галуа. Четкая формулировка теоремы Абеля еще более сложна и здесь не приводится, см. [Sk15, Замечание 7]. Экономнее решить проблему разрешимости, доказав следующую теорему Галуа (более слабую и более просто доказываемую, чем теорема Галуа 1.3). Для $X \subset \mathbb{C}$ комплексное число называется $X$-радикальным, если его можно получить из множества $X \cup\{1\}$ при помощи операций из определения радикальности. Теорема 1.4 (Галуа). Существуют такие $a_{0}, a_{1}, a_{2}, a_{3}, a_{4} \in \mathbb{C}$, что ни один корень уравнения $x^{5}+a_{4} x^{4}+\ldots+a_{1} x+a_{0}=0$ не является $\left\{a_{0}, a_{1}, \ldots, a_{4}\right\}$-радикальным. Теорема 1.5. Существует алгоритм, определяющий для данных $a_{n-1}, \ldots, a_{0} \in \mathbb{Q}$, все ли корни уравнения $x^{n}+a_{n-1} x^{n-1}+\ldots+$ $a_{1} x+a_{0}=0$ радикальны. Теорема 1.5 доказывается при помощи критерия разрешимости Галуа 2.13.b и оценки на число операций. ## 1.4 План Этот проект распадается на три формально независимых куска (в первых двух используется определение радикальности из п. 2.2). (1) В п. 2.3 обсуждается метод (резольвент Лагранжа) решения уравнений. Формально он не используется в доказательствах неразрешимости. Однако знакомство с ним будет полезно, поскольку доказательства неразрешимости были придуманы при анализе этого метода, поскольку это знакомство поможет контролировать правильность промежуточных гипотез, возникающих при доказательствах неразрешимости, и поскольку этот метод нужен для доказательства теоремы 1.5 . (2) Доказательство теоремы Руффини 2.7 основано на идее симметрии и намечено в п. 3.1. К нему подводит п. 2.5. П. 2.4 подводит и к п. 2.5, и к доказательству теоремы Галуа 1.4. (3) Доказательство теоремы 1.2 о разрешимости в вещественных радикалах основано на идее сопряжения (или алгебраической симметрии). $\mathrm{K}$ нему подводят п. 2.1, 3.2 и 3.3. Теоремы 1.3 и 1.5 не доказываются в этом тексте, см. доказательство первой в [ZSS, §5], [Sk19, §9]. Теорема Галуа 1.4 доказывается в дополнительных задачах, cp. [Sk15, Sk19], при помощи редукции к теореме Руффини 2.7 , использующей идею сопряжения (п. 2.1, 3.2 и 3.3$)$. ## 1.5 Рекомендации участникам Участник (или группа участников) конференции, решающий задачи проекта, получает «боб» за каждое записанное решение, оцененное в «+» или «+.». Дополнительные бобы могут выдаваться за красивые решения, решения сложных проблем, или оформление некоторых решений в системе $\mathrm{T}_{\mathrm{E}} \mathrm{X}$. У жюри бесконечно много бобов. Решения можно сдавать и устно, отдавая один боб за каждые пять попыток (неважно, удачных или нет). Если задача выделена словом «теорема» («лемма», «следствие» и т. д.) и жирным шрифтом, то её утверждение более важное. Как правило, мы приводим (в виде задачи) формулировку красивого или важного утверждения перед его доказательством. В таких случаях для доказательства утверждения могут потребоваться последующие задачи. Если Вы застряли на какой-то другой задаче, также перейдите к следующим, они могут помочь. Приглашаем Вас обсуждать с жюри возникающие вопросы. Особо успешным решателям мы выдаем дополнительные задачи для исследования. Пожалуйста, сообщите нам, если Вы знаете решения каких-то из предложенных задач. Если Вы подтвердите свои знания, сообщив нам решения некоторых из них, Вам будет разрешено не получать плюсы по всем этим задачам, но пользоваться ими при решении остальных. ## 2 Задачи до промежуточного финиша В этом тексте равенства, включающие многочлен $f$ (или $f_{j}$ ) означают равенство многочленов (покоэффициентное). В п. 2.1, 3.2 и 3.3 «многочлен с рациональными коэффициентами» коротко называется многочленом. Обозначим $$ \varepsilon_{q}:=\cos (2 \pi / q)+i \sin (2 \pi / q) $$ ## 2.1 Одно извлечение квадратного корня 2.1. Представимо ли следующее число в виде $a+\sqrt{b}$, где $a, b \in \mathbb{Q}$ : (a) $\sqrt{3+2 \sqrt{2}}$; (b) $\frac{1}{7+5 \sqrt{2}}$; (c) $\sqrt[3]{7+5 \sqrt{2}}$ (d) $\cos (2 \pi / 5)$; (e) $\sqrt[3]{2} ; \quad$ (f) $\sqrt{2}+\sqrt[3]{2}$; (g) $\cos (2 \pi / 9)$; (h) ${ }^{*} \sqrt{2+\sqrt{2}}$; (i) ${ }^{*} \cos (2 \pi / 7)$; (j) $\sqrt{2}+\sqrt{3}+\sqrt{5}$. Лемма 2.2. Пусть $r \in \mathbb{R}-\mathbb{Q}$ и $r^{2} \in \mathbb{Q}$. (a) О неприводимости. Многочлен $x^{2}-r^{2}$ неприводим над $\mathbb{Q}$. (b) О линейной независимости. Если $a, b \in \mathbb{Q}$ и $a+b r=0$, то $a=b=0$. (c) Если многочлен имеет корень $r$, то этот многочлен делится на $x^{2}-r^{2}$. (d) $\mathbf{O}$ сопряжении. Если многочлен имеет корень $r$, то корнем этого многочлена является также число $-r$. (е) О сопряжении. Если $a, b \in \mathbb{Q}$ и многочлен имеет корень $a+b r$, то корнем этого многочлена является также число $a-b r$. (f) Если $a, b \in \mathbb{Q}$ и кубический многочлен имеет корень $a+b r$, то он имеет рациональный корень. Теорема 2.3. Если многочлен степени выше второй неприводим над $\mathbb{Q}$, то ни один из его корней не представим в виде $a \pm \sqrt{b}$, где $a, b \in \mathbb{Q}$. Лемма 2.4 (о расширении). Пусть число можно получить из числа 1 при помощи нескольких операций сложений, вычитаний, умножений, делений на ненулевые числа, и одной операции извлечения квадратного корня из положительного числа (т.е. число вещественно построимо с извлечением корня только один раз). Тогда оно имеет вид $a \pm \sqrt{b}$, где $a, b \in \mathbb{Q}$ и $b>0$. 2.5.* Для каких $n$ число $\cos (2 \pi / n)$ представимо в виде $a+\sqrt{b}$, где $a, b \in \mathbb{Q}$ ? ## Подсказки к п. 2.1 2.2. (a) Если многочлен $x^{2}-r^{2}$ приводим над $\mathbb{Q}$, то он имеет рациональный корень. Противоречие. (b) Если $b \neq 0$, то $r=-a / b \in \mathbb{Q}$, что невозможно. Поэтому $b=0$, а значит, $a=0$. (c) Поделим многочлен с остатком ${ }^{3}$ на $x^{2}-r^{2}$ : $$ P(x)=\left(x^{2}-r^{2}\right) Q(x)+m x+n $$ Подставляя $x=r$, по лемме о линейной независимости (см. п. (b)) получаем, что остаток нулевой. (d) Из п. (с) следует, что если $R^{2}=r^{2}$, то $R$ есть корень многочлена. Указание к другому решению. Отображение $u \mapsto \bar{u}$ множества $\mathbb{Q}[\sqrt{2}]:=\{a+b \sqrt{2}: a, b \in \mathbb{Q}\}$ в себя корректно определено формулой $\overline{a+b r}:=a-b r$. Кроме того, $\overline{u+v}=\bar{u}+\bar{v}$ и $\overline{u \cdot v}=\bar{u} \cdot \bar{v}$ для любых $u, v \in \mathbb{Q}[\sqrt{2}]$.[^8](е) Обозначим через $P$ многочлен из условия, и пусть $G(t):=P(a+$ $b t)$. Тогда $G(r)=0$. Значит, по пункту (d) имеем $G(-r)=0$. (f) Если $b=0$, то утверждение доказано. В противном случае по п. (е) многочлен имеет (различные) корни $a \pm b r$, значит третий корень рационален по теореме Виета. 2.4. Было бы достаточно доказать, что множество чисел такого вида замкнуто относительно сложения, вычитания, умножения и деления. Это, естественно, не так: $(1+\sqrt{2})+(1+\sqrt{3})$ не представимо в виде $a \pm \sqrt{b}$, где $a, b \in \mathbb{Q}$ (докажите!). ## 2.2 Определение радикальности многочлена Решение квадратного уравнения $t^{2}+b t+c=0$ можно выразить формулами $$ (x-y)^{2}=(x+y)^{2}-4 x y=b^{2}-4 c \text { и } x=\frac{x+y+(x-y)}{2}=\frac{-b+(x-y)}{2} $$ Эти формулы показывают, что корень $x$ квадратного уравнения выразим в радикалах (в смысле, строго определенном ниже) через коэффициенты $-b=x+y, c=x y$ квадратного уравнения. Обозначим элементарные симметрические многочлены $$ \sigma_{1}\left(x_{1}, \ldots, x_{n}\right):=x_{1}+\ldots+x_{n}, \quad \ldots, \quad \sigma_{n}\left(x_{1}, \ldots, x_{n}\right)=x_{1} \cdot \ldots \cdot x_{n} $$ Если число $n$ и аргументы $x_{1}, \ldots, x_{n}$ ясны из контекста, то они пропускаются из обозначений. Многочлен $p \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ называется (комплексно) радикальным если $p$ можно добавить в набор $\left\{\sigma_{1}, \ldots, \sigma_{n}\right\} \cup \mathbb{C}$ многочленов цепочкой операций следующего вида: - добавить в набор сумму или произведение уже имеющихся многочленов; - если многочлен из набора равен $f^{k}$ для некоторых $f \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ и целого $k>1$, то добавить в набор многочлен $f$. Замечание 2.6. (а) Например, к многочленам $x^{2}+2 y$ и $x-y^{3}$ операциями первого типа можно добавить многочлен $-5\left(x^{2}+2 y\right)^{2}+$ $3\left(x^{2}+2 y\right)\left(x-y^{3}\right)^{6}$. А к многочлену $x^{2}-2 x y+y^{2}$ операцией второго типа можно добавить многочлен $x-y$ (или $y-x$ ). (b) Операции первого типа добавляют многочлен с комплексными коэффициентами от уже имеющихся. (c) По теореме Виета $\sigma_{1}, \ldots, \sigma_{n}$ есть коэффициенты многочлена $$ t^{n}-\sigma_{1} t^{n-1}+\ldots+(-1)^{n-1} \sigma_{n-1} t+(-1)^{n} \sigma_{n} \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right][t] $$ с корнями $x_{1}, \ldots, x_{n}$. Поэтому радикальность многочлена $x_{1}$ равносильна выразимости (в указанном смысле) через коэффициенты этого многочлена его корня $x_{1}$. (d) Радикальность многочлена $x_{1}$ равносильна существованию таких - целых положительных чисел $s, k_{1}, \ldots, k_{s}$, - многочленов $f_{1}, \ldots, f_{s}$ от $n$ переменных и $p_{0}, p_{1}, \ldots, p_{s}$ от $n, n+$ $1, \ldots, n+s$ переменных, соответственно, с комплексными коэффициентами, что $$ \left\{\begin{array}{l} f_{1}^{k_{1}}=p_{0}\left(\sigma_{1}, \ldots, \sigma_{n}\right) \\ f_{2}^{k_{2}}=p_{1}\left(\sigma_{1}, \ldots, \sigma_{n}, f_{1}\right) \\ \cdots \\ f_{s}^{k_{s}}=p_{s-1}\left(\sigma_{1}, \ldots, \sigma_{n}, f_{1}, \ldots, f_{s-1}\right) \\ x_{1}=p_{s}\left(\sigma_{1}, \ldots, \sigma_{n}, f_{1}, \ldots, f_{s}\right) \end{array}\right. $$ В этих равенствах мы опускаем переменные $\left(x_{1}, \ldots, x_{n}\right)$ многочленов $\sigma_{1}, \ldots, \sigma_{n}, f_{1}, \ldots, f_{s}$. (е) Всегда ли можно, зная $x+y$ и $x y$, однозначно найти $x$ ? Вот простейшая формализация этого вопроса: существует ли отображение $f: \mathbb{R}^{2} \rightarrow \mathbb{R}$, для которого $f(x+y, x y)=x$ при любых $x, y \in \mathbb{R}$ ? Ответ: не существует (действительно, рассмотрите пары $x=1, y=2$ и $x=2, y=1$ ). Итак, радикальность не дает «нахождения»в указанном выше смысле. Аналогично, зная $\sigma_{1}=x+y+z, \sigma_{2}=x y+y z+z x$ и $\sigma_{3}=x y z$, невозможно однозначно найти $(x-y)(y-z)(z-x)$ (действительно, рассмотрите тройки $x=0, y=1, z=-1$ и $x=0, y=-1, z=1)$. Теорема 2.7 (Руффини). Ни для какого $n \geqslant 5$ многочлен $x_{1}$ не радикален. Из доказательства будет вытекать, что даже многочлен $x_{1} x_{2}+$ $x_{2} x_{3}+x_{3} x_{4}+x_{4} x_{5}+x_{5} x_{1}$ не радикален для $n=5$. ## 2.3 Решение уравнений малых степеней 2.8. Какие из следующих многочленов радикальны для $n=3$ ? (a) $(x-y)(y-z)(z-x)$; (b) $x^{9} y+y^{9} z+z^{9} x$; (c) $x$. В задаче 2.8 и далее используйте основную теорему о симметрических многочленах, см., например, [ZSS, 4.6.3с]. Подсказкой к п. (с) являются следующие задачи 2.9.а и 2.11.c. 2.9. Многочлен $f \in \mathbb{R}\left[u_{1}, u_{2}, \ldots, u_{n}\right]$ называется циклически симметрическим, если $f\left(u_{1}, u_{2}, \ldots, u_{n}\right)=f\left(u_{2}, u_{3}, \ldots, u_{n-1}, u_{n}, u_{1}\right)$. (a) Найдите хотя бы одну пару $\alpha, \beta \in \mathbb{C}$, для которой многочлен $(u+v \alpha+w \beta)^{3}$ циклически симметрический, а многочлен $u+v \alpha+w \beta$ - нет. (b) Получите многочлен $x_{1} x_{3}+x_{3} x_{5}+x_{5} x_{7}+x_{7} x_{9}+x_{9} x_{1}$ операциями из определения радикальности из некоторых циклически симметрических многочленов от $x_{1}, x_{2}, \ldots, x_{10}$. 2.10. Какие из следующих многочленов радикальны для $n=4$ ? (a) $(x-y)(x-z)(x-t)(y-z)(y-t)(z-t)$; (b) $x y+z t$; (c) $x+y-z-t$; (d) $x$. 2.11. Решите системы уравнений $(x, y, z, t$ - неизвестные, $a, b, c, d$ известны ): (a) $\left\{\begin{array}{l}x+y+z+t=a, \\ x+y-z-t=b, \\ x-y+z-t=c, \\ x-y-z+t=d ;\end{array} \quad\right.$ (b) $\left\{\begin{array}{l}x+y+z+t=a, \\ x+i y-z-i t=b, \\ x-y+z-t=c, \\ x-i y-z+i t=d ;\end{array}\right.$ (c) $\left\{\begin{array}{l}x+y+z=a, \\ x+\varepsilon_{3} y+\varepsilon_{3}^{2} z=b, \\ x+\varepsilon_{3}^{2} y+\varepsilon_{3} z=c .\end{array}\right.$ Выражения из задачи 2.11 называются резолъвентами Лагранжа. Они «лучше»корней, поскольку «симметричнее» в следующем смысле. Решение кубического уравнения при помощи резолъвент Лагранжа (решение задачи 2.8 (c)). Для нахождения корней $x, y, z$ кубического уравнения достаточно найти выражения $a, b, c$ из задачи 2.11 (c). (Заметим, что метод дель Ферро из задачи [ZSS, 4.2.2] фактически приводит к тому же.) По теореме Виета $a=a(x, y, z)-$ коэффициент уравнения. При замене $x \leftrightarrow y$ многочлен $b=b(x, y, z)$ переходит в $\varepsilon_{3} c$, а $c=c(x, y, z)$ в $\varepsilon_{3}^{2} b$ (проверьте!). Значит, многочлены $b c$ и $b^{3}+c^{3}$ не меняются при этой замене. Аналогично они не меняются при замене $z \leftrightarrow y$. Поэтому многочлены $b c$ и $b^{3}+c^{3}$ симметрические, т. е. не меняются при любой перестановке переменных. Тогда из теоремы Виета и теоремы о представимости симметрического многочлена в виде многочлена от элементарных симметрических многочленов (утверждение [ZSS, 4.6.3с]) следует, что эти многочлены от $x, y, z$ представляются в виде многочленов от коэффициентов уравнения. Теперь, решая квадратное уравнение, можно получить $b^{3}$ и $c^{3}$. Далее легко получить сами $b$ и $c$. Ввиду теоремы Руффини 2.7 метод резольвент Лагранжа, продемонстрированный на примере решения уравнений 3 -й и 4 -й степени (задачи 2.8 (c) и $2.10(\mathrm{~d})$ ), не работает для уравнения 5 -й степени. Сообразите, почему! Обозначим через $\Sigma_{q}$ множество перестановок $q$-элементного множества. For a permutation $\alpha \in \Sigma_{q}$ denote $$ \vec{u}_{\alpha}:=\left(u_{\alpha(1)}, \ldots, u_{\alpha(q)}\right) $$ Определим резольвенту Лагранжа как $$ t\left(u_{1}, \ldots, u_{q}\right):=\varepsilon_{q} u_{1}+\varepsilon_{q}^{2} u_{2}+\ldots+\varepsilon_{q}^{q} u_{q} $$ Определим резолъвенту Галуа как $$ Q\left(u_{1}, \ldots, u_{q}, y\right):=\prod_{\alpha \in \Sigma_{q}}\left(y-t\left(\vec{u}_{\alpha}\right)\right) \in \mathbb{Q}\left[\varepsilon_{q}\right]\left[u_{1}, \ldots, u_{q}, y\right] $$ 2.12. (a) Имеем $Q\left(\varepsilon_{q} u_{1}, \ldots, \varepsilon_{q} u_{q}, y\right)=Q\left(u_{1}, \ldots, u_{q}, y\right)$. (b) Для некоторого $R_{Q} \in \mathbb{Q}\left[\varepsilon_{q}\right][z]$ имеем $Q\left(u_{1}, \ldots, u_{q}, y\right)=$ $=R_{Q}\left(u_{1}, \ldots, u_{q}, y^{q}\right)$. (c) Если $x_{1}, \ldots, x_{5}$ - корни многочлена $f \in \mathbb{Q}[x] 5$-й степени, то $Q\left(x_{1}, \ldots, x_{5}, y\right) \in \mathbb{Q}\left[\varepsilon_{5}\right][y]$ и даже $Q\left(x_{1}, \ldots, x_{5}, y\right) \in \mathbb{Q}[y]$. Многочлен $R_{Q}\left(x_{1}, \ldots, x_{5}, z\right) \in \mathbb{Q}[z]$ называется разрешающим многочленом для $f$. (d)* Все корни разрешающего многочлена для $f(x)=x^{5}+15 x+$ 11 (а значит, и самого многочлена $f$ ) радикальны. Теорема 2.13.* (а) При $a, b \in \mathbb{R}$ все корни уравнения $x^{5}+a x+$ $b=0$ радикальны тогда и только тогда, когда $a=\frac{15 \pm 20 c}{c^{2}+1}$ и $b=$ $44 \mp 8 c$ $\frac{44+8 c}{c^{2}+1}$ для некоторого $c \in \mathbb{Q}, c \geqslant 0$. (b) (Критерий Галуа разрешимости) Для любых $a_{n-1}, \ldots, a_{0} \in$ $\mathbb{Q}$ все корни уравнения $A(x):=x^{n}+a_{n-1} x^{n-1}+\ldots+a_{1} x+a_{0}=0$ радикальны тогда и только тогда, когда некоторый набор многочленов степени 1 с коэффициентами в $\mathbb{Q}$ может быть получен из $\{A\}$ при помощи следующих операций: - (факторизация) если один из многочленов равен $P_{1} P_{2}$ для некоторых $P_{1}, P_{2} \in \mathbb{Q}[x]$, не являющихся константами, то заменим $P_{1} P_{2}$ на $P_{1}$ и $P_{2} ;$ - (извлечение корня) если один из наших многочленов равен $P\left(x^{q}\right)$ для некоторого $P \in \mathbb{Q}[x]$, то заменим $P\left(x^{q}\right)$ на $P(x)$; - (взятие резольвенты Галуа) заменим один из наших многочленов $P$ на многочлен $Q\left(y_{1}, \ldots, y_{q}, y\right)$, где $y_{1}, \ldots, y_{q}$ - все корни многочлена $P$. (По задаче 2.12.c $Q\left(y_{1}, \ldots, y_{q}, y\right) \in \mathbb{Q}[y]$.) Часть (a) выводится из (b) [PSo]. Часть «тогда» в (b) проще и доказывается методом резольвент Лагранжа, разобранным в этом пункте. Часть «только тогда» в (b) сложнее и доказывается аналогично теоремам Галуа 1.3, 1.4. ## 2.4 Единственность способа решения квадратного уравнения Системы уравнений из этого и следующего пунктов возникают при решении уравнений в радикалах («при помощи одного радикала»), см. замечание 2.6.d. 2.14. (a,b) Решите систему уравнений в многочленах $f(x, y)$, $p(u, v)$ и $q(u, v, w)$ с вещественными коэффицентами: (a) $\left\{\begin{array}{l}f^{2}(x, y)=p(x+y, x y) \\ x=q(x+y, x y, f(x, y))\end{array}\right.$. (b) $\left\{\begin{array}{l}f^{k}(x, y)=p(x+y, x y) \\ x=q(x+y, x y, f(x, y))\end{array}\right.$, где $k>0$ целое. $\left(\mathrm{c}, \mathrm{d}^{*}\right)$ Решите аналоги п. (a,b) с заменой многочлена $f$ на функцию $f: \mathbb{R}^{2} \rightarrow \mathbb{R}$ (не предполагаемую непрерывной). Системе уравнений из 2.14.а удовлетворяют, например, многочлены $$ f(x, y)=x-y, \quad p(u, v)=u^{2}-4 v \quad \text { и } \quad q(u, v, w)=\frac{u+w}{2} $$ 2.15. Пусть $f, g \in \mathbb{R}[x, y]$. (a) Лемма. Если $f g=0$, то $f=0$ или $g=0$. Предостережения: существуют функции $F, G: \mathbb{R} \rightarrow \mathbb{R}$, для которых $F G=0, F \neq 0, G \neq 0$; существуют два разных многочлена от двух переменных, равные в бесконечном множестве точек; не пользуйтесь без доказательства тем, что если значения многочленов от двух переменных совпадают в любой точке, то эти многочлены равны. (b) Если $f^{2}=g^{2}$, то $f=g$ или $f=-g$. (c) Если $f^{2}+f g+g^{2}=0$, то $f=0$ и $g=0$. (d) Если $f^{3}=g^{3}$, то $f=g$. (е) Если $f^{5}=g^{5}$, то $f=g$. (f) $f^{5}-g^{5}=(f-g)\left(f-\varepsilon_{5} g\right)\left(f-\varepsilon_{5}^{2} g\right)\left(f-\varepsilon_{5}^{3} g\right)\left(f-\varepsilon_{5}^{4} g\right)$. Для доказательства утверждений 2.14.bd полезны следующие понятия и лемма. Многочлен $f$ от двух переменных $x, y$ называется симметрическим, если $f(x, y)=f(y, x)$, и антисимметрическим, если $f(x, y)=$ $-f(y, x)$. 2.16. (а) Лемма. Если $f \in \mathbb{R}[x, y]$ - многочлен с вещественными коэффициентами от двух переменных и многочлен $f^{2}$ симметрический, то $f$ либо симметрический, либо антисимметрический. (b) Лемма. Если $f \in \mathbb{R}[x, y]$ и многочлен $f^{2 k+1}$ симметрический, то $f$ симметрический. (c) Если $f \in \mathbb{R}[x, y]$ антисимметрический, то существует симметрический многочлен $a \in \mathbb{R}[x, y]$, для которого $f=(x-y) a$. Для доказательства полезна лемма 2.15.а, очень полезная и при решении других задач. 2.17. Для каких из утверждений 2.15 и 2.16 справедливы аналоги для многочленов с комплексными коэффициентами? Вот обобщение утверждения 2.14 на любое количество шагов из определения радикальности (п. 2.2). 2.18. Рациональной функиией называется «формальное отношение многочленов», т.е. пара $f / g:=(f, g)$ многочленов, в которой $g \neq 0$, с точностью до следующей эквивалентности: $f / g \sim f^{\prime} / g^{\prime}$ при $f g^{\prime}=f^{\prime} g$. При этом многочлен $f$ отождествляется с парой $(f, 1)$. (a) Дайте определения суммы и произведения рациональных функций. Проверьте их корректность. (b) Возьмем систему из замечания 2.6.(d) для $n=2$, в которой $f_{j}$ и $p_{j}$ рациональные функции, а не обязательно многочлены, и которая минимальна, т.е. нет системы с меньшим $s$ и $f_{j}^{k}$ не представляется в виде рациональной функции от $x+y, x y, f_{1}, \ldots, f_{j-1}$ ни для каких $j=1, \ldots, s$ и $k0$. Для доказательства полезны следующие понятие и утверждение. Многочлен $f \in \mathbb{R}[x, y, z]$ называется циклически симметрическим, если $f(x, y, z)=f(y, z, x)$. 2.20. Если $f \in \mathbb{R}[x, y, z]$ и многочлен (a) $f^{3}$; (b) $f^{2}$ циклически симметрический, то $f$ циклически симметрический. Замечание 2.21 (ср. с решением задачи 2.8.c). Не существует таких многочленов $f_{1}(x, y, z), \quad f_{2}(x, y, z), \quad p_{0}(u, v, w), \quad p_{1}\left(u, v, w, \tau_{1}\right), \quad p_{2}\left(u, v, w, \tau_{1}, \tau_{2}\right)$ с вещественными коэффициентами, для которых $$ \left\{\begin{array}{l} f_{1}^{2}=p_{0}\left(\sigma_{1}, \sigma_{2}, \sigma_{3}\right) \\ f_{2}^{3}=p_{1}\left(\sigma_{1}, \sigma_{2}, \sigma_{3}, f_{1}\right) \\ x=p_{2}\left(\sigma_{1}, \sigma_{2}, \sigma_{3}, f_{1}, f_{2}\right) \end{array}\right. $$ Обобщение замечания 2.21 на любое количество шагов формализуется определением вещественной радикальности, которое получается из его комплексного аналога (§2.3) заменой комплексных коэффициентов на вещественные. Формулы в начале п. 2.2 показывают, что многочлен $x$ вещественно радикален для $n=2$. Решение задачи 2.8.аь показывает, что оба многочлена $$ (x-y)(y-z)(z-x) \quad \text { и } \quad x^{9} y+y^{9} z+z^{9} x $$ вещественно радикальны для $n=3$. Теорема 2.22. Многочлен $x$ не является вещественно радикальным для $n=3$. Теорема 2.22 есть еще одна формализация того, что корень кубического уравнения не выразим в вещественных радикалах через его коэффициенты, ср. с замечанием 1.1.е. Она вытекает из следующей леммы. Лемма 2.23 (о сохранении циклической симметричности). Если $q>0$ целое, $f \in \mathbb{R}[x, y, z]$ и многочлен $f^{q}$ циклически симметрический, то $f$ циклически симметрический. 2.24. Аналоги каких утверждений этого пункта справедливы для многочленов с комплексными коэффициентами? K алгоритмам решения алгебраических уравнений представляют Б. Вукорепа, А. Глебов, А. Еннэ, А. Скопенков, А. Чиликов ## 3 Задачи после промежуточного финиша ## 3.1 Неразрешимость «в многочленах» Определение радикальности многочлена приведено в п. 2.2. Формально, теорема Руффини 2.7 вытекает из леммы 3.4. Самое трудное и интересное - придумать формулировку этой леммы. Для этого докажем следующие более простые факты. Сообразите, почему многочлен $x$ не является многочленом от $x+y$ и $x y$. 3.1. Многочлен $x_{1}$ не радикален для $n=3$ так, что вторая операция из определения радикальности применяется только для (а) $k=2$ (подсказка: см. задачу 2.24 ); (b) $k=3$. 3.2. Какие из следующих утверждений верны для любого $f \in$ $\mathbb{C}\left[x_{1}, \ldots, x_{5}\right]$ ? (a) Если $f^{3}$ циклически симметрический, то $f$ циклически симметрический. (b) Если $f^{5}$ циклически симметрический, то $f$ циклически симметрический. (c) Если $f^{3}$ симметрический, то $f$ симметрический. (d) Если $f^{2}$ симметрический, то $f$ симметрический. Циклом длины 3 называется перестановка $n$-элементного множества, переставляющая некоторые 3 элемента по циклу и оставляющая на месте каждый из оставшихся элементов. Многочлен $f \in$ $\mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ называется четносимметрическим, если для любого цикла $\quad \alpha$ длины 3 многочлены $f\left(x_{1}, x_{2}, \ldots, x_{n}\right) \quad$ и $f\left(x_{\alpha(1)}, x_{\alpha(2)}, \ldots, x_{\alpha(n)}\right)$ равны. 3.3. (а) Придумайте циклически симметрический многочлен, не являющийся четносимметрическим. (b) Если перестановка переводит в себя многочлен, построенный Вами в решении задачи 3.2.d, то она представляется в виде композиции циклов длины 3. Лемма 3.4 (о сохранении четносимметричности). Если $q>0$ целое, $f \in \mathbb{C}\left[x_{1}, \ldots, x_{5}\right]$ и многочлен $f^{q}$ четносимметрический, то $f$ четносимметрический. 3.5. Пусть $f \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ - многочлен. (a) Если многочлен $f^{7}$ четносимметрический, то $f$ четносимметрический. (b) Если $n \geqslant 5$ и многочлен $f^{3}$ четносимметрический, то $f$ четносимметрический. (c) Если $n \geqslant 5$, то любой цикл длины 3 на $n$-элементном множестве разлагается в произведение перестановок вида $(a b)(c d)$ с различными $a, b, c, d$ (т.е. в произведение композиций транспозиций с непересекающимися носителями). 3.6. Определение рациональной вещественной (комплексной) радикальности аналогично определению радикальности, только вместо многочленов берутся рациональные функции (с соответствующими коэффициентами; см. определение в задаче 2.18). Является ли многочлен $x_{1}$ (a) вещественно рационально радикальным для $n=3$ ? (b) (комплексно) рационально радикальным для $n=5$ ? ## 3.2 Одно извлечение корня третьей степени Здесь развиваются идеи из п. 2.1. 3.7. Представимо ли следующее число в виде $a+b \sqrt[3]{2}+c \sqrt[3]{4}$, где $a, b, c \in \mathbb{Q}:$ (a) $\sqrt{3}$; (b) $\frac{1}{1+5 \sqrt[3]{2}+\sqrt[3]{4}}$; (c) $\cos (2 \pi / 9)$; (d) $\sqrt[5]{3}$; (e) $\sqrt[3]{3}$; (f) наибольший вещественный корень многочлена $x^{3}-4 x+2$; (g)* единственный вещественный корень многочлена $x^{3}-6 x-6$; (h)* единственный вещественный корень многочлена $x^{3}-9 x-12$ ? Лемма 3.8. Пусть $r \in \mathbb{R}-\mathbb{Q}$ и $r^{3} \in \mathbb{Q}$. (a) О неприводимости. Многочлен $x^{3}-r^{3}$ неприводим над $\mathbb{Q}$. (b) О линейной независимости. Если $a, b, c \in \mathbb{Q}$ и $a+b r+c r^{2}=$ 0 , то $a=b=c=0$. $\left(\mathrm{b}^{\prime}\right)$ О линейной независимости над $\mathbb{Q}\left[\varepsilon_{3}\right]$. Если $$ k, l, m \in \mathbb{Q}\left[\varepsilon_{3}\right]:=\left\{u+v \varepsilon_{3}: u, v \in \mathbb{Q}\right\} $$ и $k+l r+m r^{2}=0$, то $k=l=m=0$. (c) Если многочлен имеет корень $r$, то этот многочлен делится на $x^{3}-r^{3}$. (d) О сопряжении. Если многочлен имеет корень $r$, то корнями этого многочлена являются также числа $\varepsilon_{3} r$ и $\varepsilon_{3}^{2} r$. (е) О сопряжении. Если $a, b, c \in \mathbb{Q}$ и многочлен имеет корень $x_{0}:=a+b r+c r^{2}$, то корнями этого многочлена являются также числа $$ x_{1}:=a+b \varepsilon_{3} r+c \varepsilon_{3}^{2} r^{2} \quad \text { и } \quad x_{2}:=a+b \varepsilon_{3}^{2} r+c \varepsilon_{3} r^{2} $$ (f) $\mathbf{O}$ рациональности. Если $a, b, c \in \mathbb{Q}$, то число $a+b r+c r^{2}$ является корнем некоторого ненулевого многочлена степени 3 . Теорема 3.9. Если многочлен неприводим над $\mathbb{Q}$ и имеет корень вида $a+b r+c r^{2} \notin \mathbb{Q}$, где $r \in \mathbb{R}-\mathbb{Q}$ и $a, b, c, r^{3} \in \mathbb{Q}$, то степень многочлена равна 3 и он имеет ровно один вещественный корень. Лемма 3.10 (о расширении). Число, вещественно радикальное с извлечением корня только один раз, причём третьей степени, имеет вид $a+b r+c r^{2}$, где $r \in \mathbb{R}$ и $a, b, c, r^{3} \in \mathbb{Q}$. ## 3.3 Одно извлечение корня простой степени 3.11. Представимо ли следующее число в виде $$ a_{0}+a_{1} \sqrt[7]{2}+a_{2} \sqrt[7]{2^{2}}+\ldots+a_{6} \sqrt[7]{2^{6}} $$ где $a_{0}, a_{1}, a_{2}, \ldots, a_{6} \in \mathbb{Q}$ ? (a) $\sqrt{3}$; (b) $\cos \frac{2 \pi}{21}$; (c) $\sqrt[11]{3}$; (d) $\sqrt[7]{3}$; (e) какой-нибудь из корней многочлена $x^{7}-4 x+2$. Ответы: не представимы. Доказательства аналогичны решениям задач 3.7. Используйте сформулированные ниже леммы. Лемма 3.12. Пусть $q$ простое, $r \in \mathbb{R}-\mathbb{Q}$ и $r^{q} \in \mathbb{Q}$. (a) $\mathbf{O}$ неприводимости. Многочлен $x^{q}-r^{q}$ неприводим над $\mathbb{Q}$. (b) О линейной независимости. Если $A$-многочлен степени меньше $q$ и $A(r)=0$, то $A=0$. (c) О сопряжении. Если многочлен имеет корень $r$, то он имеет также корни $r \varepsilon_{q}^{k}$ для каждого $k=1,2,3, \ldots, q-1$. (d) О рациональности. Если $A$ - многочлен, то число $A(r)$ является корнем некоторого ненулевого многочлена степени не выше $q$. Обозначим $$ \mathbb{Q}\left[\varepsilon_{q}\right]:=\left\{a_{0}+a_{1} \varepsilon_{q}+a_{2} \varepsilon_{q}^{2}+\ldots+a_{q-2} \varepsilon_{q}^{q-2}: a_{0}, \ldots, a_{q-2} \in \mathbb{Q}\right\} $$ 3.13. Пусть $q$ простое, $r \in \mathbb{C}-\mathbb{Q}\left[\varepsilon_{q}\right]$ и $r^{q} \in \mathbb{Q}\left[\varepsilon_{q}\right]$. (a) Многочлен $x^{q}-r^{q}$ неприводим над $\mathbb{Q}\left[\varepsilon_{q}\right]$. (b), (c) Докажите аналоги пунктов (b), (c) предыдущей задачи для многочлена с коэффициентами в $\mathbb{Q}\left[\varepsilon_{q}\right]$. Лемма 3.14.* Пусть $q$ простое, $r \in \mathbb{R}-\mathbb{Q}$ и $r^{q} \in \mathbb{Q}$. (a) $\mathbf{O}$ неприводимости над $\mathbb{Q}\left[\varepsilon_{q}\right]$. Многочлен $x^{q}-r^{q}$ неприводим над $\mathbb{Q}\left[\varepsilon_{q}\right]$. (b) О линейной независимости над $\mathbb{Q}\left[\varepsilon_{q}\right]$. Если $A$ - многочлен степени меньше $q$ с коэффициентами в $\mathbb{Q}\left[\varepsilon_{q}\right]$ и $A(r)=0$, то $A=0$. Теорема 3.15. Пусть многочлен неприводим над $\mathbb{Q}$ и имеет иррациональный корень $A(r)$ для некоторых многочлена $A \in \mathbb{Q}[x]$ и $r \in \mathbb{R}$, причём $r^{q} \in \mathbb{Q}$ для некоторого простого $q$. Тогда многочлен имеет степень $q$ и при $q \neq 2$ не имеет других вещественных корней. Доказательство аналогично доказательствам теорем $2.3,3.9$ и решениям задач 3.11 (abc). Используйте леммы о сопряжении 3.12 (c), о рациональности $3.12(\mathrm{~d})$ и о линейной независимости над $\mathbb{Q}\left[\varepsilon_{q}\right]$ $3.14(\mathrm{~b})$. Лемма 3.16 (о расширении). Число, вещественно радикальное с извлечением корня только один раз, равно $A(r)$ для некоторых $A \in \mathbb{Q}[x]$ и $r \in \mathbb{R}$, причём $r^{q} \in \mathbb{Q}$ для некоторого $q \in \mathbb{Z}$. Доказательство аналогично лемме 3.10 о расширении. 3.17. (a-d) Докажите аналоги утверждений задачи 3.12 с заменой $\mathbb{Q}$ на произвольное подмножество $F \subset \mathbb{R}$, замкнутое относительно операций сложения, вычитания, умножения и деления на ненулевое число (и многочленов с коэффициентами в $\mathbb{Q}$ на многочлены с коэффициентами в $F$ ). ## Решения задач до промежуточного финиша 2.1. Ответы: (a), (b), (c), (d) - да, (e), (f), (g), (h), (i) - нет. (a), (c) Имеем $\sqrt{3+2 \sqrt{2}}=\sqrt[3]{7+5 \sqrt{2}}=1+\sqrt{2}$. (b) Имеем $\frac{1}{7+5 \sqrt{2}}=\frac{7-5 \sqrt{2}}{7^{2}-2 \cdot 5^{2}}=-7+5 \sqrt{2}$. (d) Имеем $\cos (2 \pi / 5)=(\sqrt{5}-1) / 4$. (е) Пусть число $\sqrt[3]{2}$ представимо. Тогда $$ 2=(\sqrt[3]{2})^{3}=\left(a^{3}+3 a b\right)+\left(3 a^{2}+b\right) \sqrt{b} $$ Так как $3 a^{2}+b \neq 0$, то $\sqrt{b} \in \mathbb{Q}$. Значит, $\sqrt[3]{2} \in \mathbb{Q}$ - противоречие. Другой способ - аналогично теореме 2.3. (f) Набросок первого решения. Проще доказать сразу, что $\sqrt[3]{2} \neq a+p \sqrt{b}+q \sqrt{c}+r \sqrt{b c}, \quad$ ни для каких $\quad a, b, c, p, q, r \in \mathbb{Q}$. Для этого достаточно доказать, что $\sqrt[3]{2} \neq u+v \sqrt{c}$ ни для каких чисел $u, v, c \in \mathbb{Q}[\sqrt{b}]:=\{x+y \sqrt{b}: x, y \in \mathbb{Q}\}$. Идея доказательства состоит в том, что числа из $\mathbb{Q}[\sqrt{b}]$ (с фиксированным $b$ ) «ничуть не хуже» рациональных чисел, т. е. сумма, разность, произведение и частное чисел из $\mathbb{Q}[\sqrt{b}]$ тоже являются числами из $\mathbb{Q}[\sqrt{b}]$ (или, говоря научно, $\mathbb{Q}[\sqrt{b}]$ - числовое поле). Поэтому можно доказывать утверждение аналогично п. (е). Набросок второго решения. Пусть число $\sqrt{2}+\sqrt[3]{2}=a+\sqrt{b}$ представимо. Оно является корнем многочлена $P(x):=\left((x-\sqrt{2})^{3}-\right.$ $2)\left((x+\sqrt{2})^{3}-2\right)$ с рациональными коэффициентами. По п. (е) $\sqrt{2}+$ $\sqrt[3]{2} \notin \mathbb{Q}$. Значит, $\sqrt{b} \notin \mathbb{Q}$. По лемме о сопряжении 2.2 (е) для $r=\sqrt{b}$, многочлен $P$ имеет корень $a-\sqrt{b}$. Так как $\sqrt{b} \notin \mathbb{Q}$, то корни $a \pm \sqrt{b}$ различны. Но у многочлена $P$ только два вещественных корня: $\sqrt{2}+\sqrt[3]{2}$ и $-\sqrt{2}+\sqrt[3]{2}$. Поэтому $a+\sqrt{b}=\sqrt{2}+\sqrt[3]{2}$ и $a-\sqrt{b}=-\sqrt{2}+\sqrt[3]{2}$. Отсюда $\sqrt[3]{2}=a \in \mathbb{Q}$. Противоречие. (g) Пусть число $\cos (2 \pi / 9)$ представимо. По формуле косинуса тройного угла оно является корнем уравнения $4 x^{3}-3 x=-\frac{1}{2}$. По лемме 2.2 (f) это уравнение имеет рациональный корень. Противоречие. Другой способ - аналогично теореме 2.3. (h) Корнями многочлена $P(x):=\left(x^{2}-2\right)^{2}-2$ являются четыре числа $\pm \sqrt{2 \pm \sqrt{2}}$, где знаки + и - не обязательно согласованы. Все эти числа иррациональны. Значит, по теореме 2.3 достаточно доказать, что многочлен $P$ не разлагается в произведение двух квадратных трехчленов с рациональными коэффициентами. Эта неразложимость следует из того, что произведение любых двух корней многочлена $P$ иррационально. (i) (Использован текст И. Брауде-Золотарёва.) Из равенства $$ \cos (2 \pi / 7)+\cos (4 \pi / 7)+\cos (6 \pi / 7)+\ldots+\cos (14 \pi / 7)=0 $$ получаем $\cos (2 \pi / 7)+\cos (4 \pi / 7)+\cos (6 \pi / 7)=-1 / 2$. Используя формулы косинуса двойного и тройного угла, получаем, что число $\cos (2 \pi / 7)$ является корнем уравнения $8 t^{3}+4 t^{2}-4 t-1=0$. Сделав замену $u=2 t$, получим $u^{3}+u^{2}-2 u-1=0$. Это уравнение не имеет рациональных корней. Значит, уравнение $8 t^{3}+4 t^{2}-4 t-1=0$ тоже не имеет рациональных корней. Поэтому многочлен $8 t^{3}+4 t^{2}-4 t-1=0$ неприводим над $\mathbb{Q}$. Теперь непредставимость вытекает из леммы $2.2(\mathrm{f})$. (j) Аналогично п. (f). 2.3. Пусть, напротив, данный многочлен $P$ имеет корень $x_{0}=$ $a \pm \sqrt{b}$. По лемме 2.2 (е) о сопряжении и аналогично ей, корнем многочлена $P$ является также число $x_{1}=a \mp \sqrt{b}$. При $b=0$ утверждение очевидно. Поэтому считаем, что $b \neq 0$. Тогда $x_{0} \neq x_{1}$. Значит, $P(x)$ делится на $(x-a)^{2}-b$. Так как $\operatorname{deg} P>2$, то многочлен $P$ приводим. Противоречие. 2.4. Обозначим через $\sqrt{c}$ число, полученное при единственном извлечении корня, где $c \in \mathbb{Q}$. Докажите, что все полученные числа имеют вид $a+b \sqrt{c}$, где $a, b \in \mathbb{Q}$. 2.5. Ответ: тогда и только тогда, когда $n \in\{1,2,3,4,5,6,8,10,12\}$. Или, эквивалентно, $\varphi(n) \in\{1,2,4\}$. 2.8. (a) $(x-y)^{2}(y-z)^{2}(z-x)^{2}$ - симметрический многочлен. (Пункт (а) можно также свести к (b).) (b) Обозначим $$ M=x^{9} y+y^{9} z+z^{9} x \quad \text { и } \quad N=y^{9} x+x^{9} z+z^{9} y $$ Тогда многочлены $M+N$ и $M N$ симметрические. Значит, они являются многочленами от элементарных симметрических многочленов $\sigma_{1}, \sigma_{2}, \sigma_{3}$. Само же $M$ выражается через $M+N$ и $M N$ по «формуле корней квадратного уравнения», см. формулы в начале п. 2.2. 2.9. (a) $x+y \varepsilon_{3}+z \varepsilon_{3}^{2}$. (b) Обозначим $$ \begin{aligned} M & =x_{1} x_{3}+x_{3} x_{5}+x_{5} x_{7}+x_{7} x_{9}+x_{9} x_{1} \\ N & =x_{2} x_{4}+x_{4} x_{6}+x_{6} x_{8}+x_{8} x_{10}+x_{10} x_{2} \end{aligned} $$ Далее аналогично задаче 2.8.b. 2.10. (a) Квадрат $(x-y)^{2}(x-z)^{2}(x-t)^{2}(y-z)^{2}(y-t)^{2}(z-t)^{2}$ симметричен, см. 2.8.a. (b) Положим $$ M=x y+z t, \quad N=x z+y t, \quad K=x t+y z $$ По 2.8.c, $M$ «выразим в радикалах при помощи многочленов» $$ M+N+K, \quad M N+M K+N K, \quad M N K $$ Аналогично решениям задач 2.8.c выше и 2.10.d ниже, эти многочлены симметрические. Поэтому $M=x y+z t$ радикален. (c) Положим $$ M=(x+y-z-t)^{2}, \quad N=(x+z-y-t)^{2}, \quad K=(x+t-y-z)^{2} $$ Повторяя решение пункта (b), получим $M=(x+y-z-t)^{2}$. Теперь легко получить и $x+y-z-t$. Решение уравнения 4-й степени при помощи резольвент Лагранжа (решение задачи 2.10.d). Для нахождения корней $x, y, z, t$ уравнения 4 -й степени достаточно найти выражения $a, b, c, d$ от корней из задачи 2.11.а. По теореме Виета $a$-коэффициент уравнения. При замене $x \leftrightarrow y$ многочлены $c^{2}$ и $d^{2}$ меняются местами, а многочлен $b^{2}$ переходит в себя. При циклической замене $x \rightarrow y \rightarrow z \rightarrow$ $t \rightarrow x$ многочлены $b^{2}$ и $d^{2}$ меняются местами, а многочлен $c^{2}$ переходит в себя. Значит, многочлены $b^{2}, c^{2}, d^{2}$ переставляются при любой перестановке переменных. Поэтому виетовские многочлены от них, т.е. $$ b^{2}+c^{2}+d^{2}, \quad b^{2} c^{2}+b^{2} d^{2}+c^{2} d^{2}, \quad b^{2} c^{2} d^{2} $$ симметрические. Тогда эти многочлены от $x, y, z$ представляются в виде многочленов от коэффициентов уравнения. Теперь, решая кубическое уравнение, можно получить сами $b^{2}, c^{2}, d^{2}$. Далее легко получить $b, c, d$. 2.11. Используйте равенства $1+\varepsilon+\varepsilon^{2}=0$ и $1+i+i^{2}+i^{3}=0$. 2.12. Для наглядности приведем решения при $q=5$. (a) Имеем $$ t\left(\varepsilon_{5} \vec{u}_{\alpha}\right)=t\left(u_{\alpha(5)}, u_{\alpha(1)}, u_{\alpha(2)}, u_{\alpha(3)}, u_{\alpha(4)}\right)=t\left(\vec{u}_{\alpha \circ(54321)}\right) $$ Следовательно, $$ \begin{aligned} & Q\left(\varepsilon_{5} u_{1}, \ldots, \varepsilon_{5} u_{5}, y\right)=\prod_{\alpha \in \Sigma_{5}}\left(y-t\left(\varepsilon_{5} \vec{u}_{\alpha}\right)\right)= \\ & =\prod_{\alpha \in \Sigma_{5}}\left(y-t\left(\vec{u}_{\alpha \circ(54321)}\right)=Q\left(u_{1}, \ldots, u_{5}, y\right)\right. \end{aligned} $$ Здесь - $(54321) \in \Sigma_{5}$ - это цикл, который отправляет 5 в 4,4 в $3, \ldots$, 1 в 5 . - последнее равенство справедливо, потому что когда $\alpha$ пробегает $\Sigma_{5}$, то же делает и $\alpha \circ(54321)$. (b) Для каждого $k=0,1,2, \ldots, 120$ найдётся однородный многочлен $P_{k} \in \mathbb{Q}\left[\varepsilon_{5}\right]\left[u_{1}, \ldots, u_{5}\right]$ («степени» $120-k$ ) такой, что коэффициент при $y^{k}$ в $Q$ равен $P\left(u_{1}, \ldots, u_{5}\right)$, т.е. $$ Q\left(u_{1}, \ldots, u_{5}, y\right)=\sum_{k=0}^{120} P_{k}\left(u_{1}, \ldots, u_{5}\right) y^{k} $$ По (а) и из однородности имеем $$ P_{k}\left(u_{1}, \ldots, u_{5}\right)=P_{k}\left(\varepsilon_{5} u_{1}, \ldots \varepsilon_{5} u_{5}\right)=\varepsilon_{5}^{-k} P_{k}\left(u_{1}, \ldots, u_{5}\right) $$ Если $k$ не кратно 5 , то $P_{k}\left(u_{1}, \ldots, u_{5}\right)=0$, что и требовалось. (c) Многочлен $Q\left(u_{1}, \ldots, u_{5}, y\right)$ симметричен по $u_{1}, \ldots, u_{5}$. Значит, все коэффициенты ( $P_{k}$ из пункта (b)) соответствующего многочлена из $\mathbb{Q}\left[\varepsilon_{5}, u_{1}, \ldots, u_{q}\right][y]$ симметричны по $u_{1}, \ldots, u_{5}$. Теперь утверждение $Q\left(x_{1}, \ldots, x_{5}, y\right) \in \mathbb{Q}\left[\varepsilon_{5}\right][y]$ следует из основной теоремы о симметрических многочленах, формул Виета и того факта, что коэффициенты $f$ рациональны. Теперь утверждение $Q\left(x_{1}, \ldots, x_{5}, y\right) \in \mathbb{Q}[y]$ доказывается аналогично [ZSS, !] 2.14. (а) Докажем, что существует такое $\alpha \in \mathbb{R}$, что $f(x, y)=$ $\alpha(x-y)$. Так как многочлен $f^{2}=p$ симметрический, то можно считать, что многочлен $q$ линеен по третьей переменной, т.е. $q(u, v, w)=$ $a(u, v)+b(u, v) w$ для некоторых $a, b \in \mathbb{R}[u, v]$ (иначе изменим $q$, сохраняя $f, p)$. Тогда $x=a(x+y, x y)+b(x+y, x y) f(x, y)$. Первое завершение решения. Получаем $p b^{2}=f^{2} b^{2}=(x-a)^{2}=$ $(y-a)^{2}$. Отсюда по лемме 2.15.b $x-a=a-y$, так как случай $x-a=y-a$ невозможен. Значит, $a=(x+y) / 2$. Тогда $(x-y)^{2}=$ $4 f^{2} b^{2}=4 p b^{2}$. Если многочлен $p=f^{2}$ постоянный, то многочлен $b= \pm(x-y) / 2 \sqrt{p}$ не симметрический - противоречие. Поэтому многочлен $p$ не постоянный. Тогда многочлен $b$ постоянный. Значит, $2 x=2 q=x+y+2 b f$, откуда $b \neq 0$ и $f=\alpha(x-y)$ для $\alpha=1 / 2 b$. Второе завершение решения (написано с использованием текста И. Богданова). Так как многочлен $x$ не симметрический и $x=q(x+$ $y, x y, f(x, y)$ ), то многочлен $f$ не симметрический. Тогда по лемме 2.16.а $f$ антисимметрический. Значит, $y=q(x+y, x y,-f(x, y))$. Итак, $$ \begin{gathered} x=a+b f \quad \text { и } \quad y=a-b f \\ \text { где } \quad a=a(x+y, x y), \quad b=b(x+y, x y) \quad \text { и } \quad f=f(x, y) \end{gathered} $$ Тогда $x+y=2 a$ и $x y=a^{2}-b^{2} f^{2}$. Отсюда $(x-y)^{2}=4 b^{2} f^{2}$. Аналогично первому завершению решения многочлен $b$ постоянный. Значит, $f=\alpha(x-y)$ для $\alpha= \pm 1 / 2 b$. (b) Докажем, что $k$ четно и существует такое $\alpha \in \mathbb{R}$, что $f(x, y)=$ $\alpha(x-y)$. Индукция по $k$ с применением п. (а) и обобщения лемм 2.15.be, 2.16. Если $k$ нечетно, то из утверждения 2.16.b получаем, что $f$ симметрический, что противоречит равенству $x=q(x+$ $y, x y, f(x, y)$ ). Если $k=4$, то $f^{2}$ либо симметрический, либо антисимметрический. Первый случай сводится к п. (а). Во втором $f^{2}(x, y)+f^{2}(y, x)=0$. Аналогично разбирается случай произвольного четного $k$. (c) Аналогично п. (а) получаем $x=a+b f$. Поэтому $f$ - дробнорациональная функция. Тогда решение аналогично п. (a). 2.15. (а) Определите стариий член многочлена так, чтобы старший член произведения равнялся произведению старших членов сомножителей. (b) Следует из п. (a). (c) Имеем $f^{2}+f g+g^{2}=\left(f+\frac{g}{2}\right)^{2}+\frac{3}{4} g^{2}=\left(f-\varepsilon_{3} g\right)\left(f-\varepsilon_{3}^{2} g\right)$. (d) Следует из п. (c). (e) Следует из п. (f). (f) Докажите и примените теорему Безу для многочленов от $u$ с коэффициентами в $\mathbb{R}[v]$. 2.16. (а) Так как $f^{2}$ симметрический, то $f(x, y)^{2}=f(y, x)^{2}$. Отсюда по утверждению 2.15.b $f(x, y)= \pm f(y, x)$. (b) Используйте аналог утверждений 2.15.ce. (c) См. указание к 2.15.f. 2.17. Ответ: 2.15.abf, 2.16.abc. 2.22. При $n=3$ множество вещественно радикальных многочленов содержится в множестве циклически симметрических многочленов. Это утверждение доказывается при помощи индукции по количеству операций из определения радикальности. Шаг индукции вытекает из леммы 2.23 о сохранении циклической симметричности. Поскольку многочлен $x$ не является циклически симметрическим, то он не является вещественно радикальным. 2.23. Доказательство можно найти в [Sk19, п. 9.4.2]. 2.24. Omвem: 2.19 .abcd, 2.20 .b, 2.23 для всех $q$, не делящихся на 3. K алгоритмам решения алгебраических уравнений представляют Б. Вукорепа, А. Глебов, А. Еннэ, А. Скопенков, А. Чиликов ## Решения задач после промежуточного финиша 3.1. (b) Используйте аналог задачи 3.2.с для $n=3$. 3.2. Ответ: (c) - верно, (a), (b), (d) - неверно. (a) См. 2.9.a. (b) Рассмотрите многочлен $x_{1}+\varepsilon_{5} x_{2}+\varepsilon_{5}^{2} x_{3}+\varepsilon_{5}^{3} x_{4}+\varepsilon_{5}^{4} x_{5}$. (d) Рассмотрите многочлен $\prod_{i0, P(1)<0$ и $P(2)>0$, то многочлен $P$ имеет вещественный корень $x_{k}$, отличный от $x_{0}$. Имеем $\overline{\varepsilon_{7}^{k}}=\varepsilon_{7}^{-k}$. Поэтому $x_{k}=\overline{x_{k}}=x_{7-k}$. Противоречие. Второе решение. Корнями многочлена $2 P(x)+1$ являются вещественные числа $y_{k}:=\cos \frac{2(3 k+1) \pi}{21}$ при $k=0, \ldots, 6$. Одно из них, а именно $y_{2}=-1 / 2$, рационально. В следубщем абзаце мы докажем, что число $y_{0}$ иррационально. (Иначе из равенства $\varepsilon_{21}^{2}-2 y_{0} \varepsilon_{21}+1=0$ следует, что $\varepsilon_{21}=a+i \sqrt{b}$ для некоторых $a, b \in \mathbb{Q}$. Тогда и число $\varepsilon_{7}=\varepsilon_{21}^{3}$ тоже имеет такой вид. Но $\varepsilon_{7}$ является корнем неприводимого ${ }^{4}$ многочлена $1+x+\ldots+$ $x^{6}$, что противоречит аналогу теоремы 2.3 для чисел вида $a+i \sqrt{b}$.) Итак, число $y_{0}$ иррационально и является корнем многочлена $\frac{2 P(x)+1}{2 x+1}$ степени 6. Тогда по леммам о сопряжении 3.12 (c) и о линейной независимости над $\mathbb{Q}\left[\varepsilon_{q}\right] 3.14(\mathrm{~b})$ этот многочлен имеет семь попарно различных корней, что невозможно. (c) Пусть число $\sqrt[11]{3}$ представимо. Тогда по лемме о рациональности $3.12(\mathrm{~d})$ существует ненулевой многочлен степени не выше 7 с корнем $\sqrt[11]{3}$. Противоречие с неприводимостью многочлена $x^{11}-3$ над $\mathbb{Q}$. (d) Пусть число $\sqrt[7]{3}$ представимо. Аналогично п. (а) все комплексные корни многочлена $x^{7}-3$ есть $A\left(r \varepsilon \frac{1}{k}\right)$ для $k=0,1,2, \ldots, 6$. Поэтому $A(r) \varepsilon_{7}^{s}=A\left(r \varepsilon_{7}\right)$ для некоторого $s \in\{1,2,3,4,5,6\}$. Отсюда по лемме о линейной независимости над $\mathbb{Q}\left[\varepsilon_{q}\right] 3.14$ (b) $a_{k}=0$ для любого $k \neq s$. Поэтому $\sqrt[7]{3}=a_{s} r^{s}$. Противоречие. (е) Пусть какой-нибудь из корней представим. Данный многочлен $P$ не имеет рациональных корней. Тогда по лемме о сопряжении 3.12.с и лемме о линейной независимости над $\mathbb{Q}\left[\varepsilon_{q}\right]$ 3.14.b $P$ имеет попарно различные корни $x_{k}:=A\left(r \varepsilon_{7}^{k}\right)$ для $k=0,1,2, \ldots, 6$. Так как $P(0)>0, P(1)<0$ и $P(2)>0$, то $P$ имеет вещественный корень $x_{k}$, отличный от $x_{0}$. Имеем $\overline{\varepsilon_{7}^{k}}=\varepsilon_{7}^{-k}$. Поэтому $x_{k}=\overline{x_{k}}=x_{7-k}$. Противоречие. 3.12. (а) Все корни многочлена $x^{q}-r^{q}$ есть $r, r \varepsilon_{q}, r \varepsilon_{q}^{2}, \ldots, r \varepsilon_{q}^{q-1}$. Пусть он приводим над $\mathbb{Q}$. Модуль свободного члена одного из унитарных сомножителей разложения рационален и равен произведе-[^9]нию модулей некоторых $k$ из этих корней, $0\operatorname{deg} P$ получаем противоречие. При $q=\operatorname{deg} P$ из условий $q \neq 2$ и $\overline{x_{k}}=x_{q-k} \neq x_{k}$ получаем единственность вещественного корня. ## Список литературы [Al] Алексеев В. Б. Теорема Абеля. М.: Наука, 1976. [AB] Akhtyamov D., Bogdanov I. Solvability of cubic and quartic equations using one radical. http://arxiv.org/abs/1411. 4990. [Ar84] Арнольд В.И. Обыкновенные дифференциальные уравнения, М. Наука, 1984. [Dor] Dörrie H. 100 Great Problems of Elementary Mathematics: Their History and Solution. New York: Dover Publ, 1965. [E2] Edwards H. M. The construction of solvable polynomials Bull. Amer. Math. Soc. 2009. V. 46. P 397-411. Errata: Bull. Amer. Math. Soc. 46 (2009), 703-704. [Es] Esterov $A$. Galois theory for general systems of polynomial equations, https://arxiv .org/abs/1801.08260 [FT] Табачников С. Л., Фукс Д. Б. Математический дивертисмент, М.: МЦНМО, 2011. [Kа] Канунников А. Л. Начала теории Галуа: разрешимость алгебраических уравнений в радикалах. http://www.mathnet.ru/conf 1015 . [Ко17] Коган E. Множественная сложность построения правильного многоугольника, https://arxiv.org/abs/1711. 05807. [Kol] Колосов B. A. Теоремы и задачи алгебры, теории чисел и комбинаторики. М.: Гелиос, 2001. [Ler] Lerner L. Galois Theory without abstract algebra. http://arxiv.org/abs/1108.4593. [Pr07-2] Прасолов В. В. Задачи по алгебре, арифметике и анализу. М.: МЦНМО, 2007. [PSo] Прасолов В.В., Соловъев Ю.П. Эллиптические функции и алгебраические уравнения. М.: Факториал, 1997. [Saf] Сафин A. Программа для построения правильных многоугольников циркулем и линейкой (доклад на ММКШ2008). http://www.mccme.ru/mmks/dec08/Safin.pdf. [Sk10] Скопенков $A$. Базисные вложения и 13 -я проблема Гильберта // Мат. Просвещение. 2010. №14. С. 143-174; http://arxiv.org/abs/1001.4011. [Sk11] Скопенков A. Простое доказательство теоремы Абеля о неразрешимости уравнений в радикалах // Мат. Просвещение. 2011. №15. С. 113-126; http://arxiv.org/abs/1102.2100. [Sk15] Skopenkov A. A short elementary proof of the insolvability of the equation of degree 5. http://arxiv.org/abs/1508. 03317 . [Sk19] Skopenkov A. Mathematics via problems: from olympiades and math circles to a profession. Algebra. AMS, Providence, to appear. [St94] Stillwell J. Galois theory for beginners, Amer. Math. Monthly, 101 (1994), 22-27. [T] Тихомиров В. М. Абель и его великая теорема // Квант. 2003. №1. C.11-15. [Vag] Вагутен Н. Сопряжённые числа // Квант. 1980. №2. С. 2632 . [ZSS] Элементы математики в задачах: через олимпиады и кружки к профессии. Сборник под редакцией А. Заславского, А. Скопенкова и М. Скопенкова. МЦНМО, 2018. http://www.mccme.ru/circles/oim/materials/sturm.pdf. ## 4 Additional problems for successful teams 4.1. (a) Let $x, y, r \in \mathbb{R}, p, g \in \mathbb{Q}[u, v]$ and $p_{1} \in \mathbb{Q}[u, v, w]$ be such that $g(x, y) \notin \mathbb{Q}(x+y, x y)$ and $$ \left\{\begin{array}{l} r^{2}=p(x+y, x y) \\ g(x, y)=p_{1}(x+y, x y, r) \end{array}\right. $$ (cf. Problem 2.14.c). Then $r \in \mathbb{Q}(x, y)$. (b) Let $x, y, r \in \mathbb{R}, p \in \mathbb{Q}[\sqrt{2}][u, v], g \in \mathbb{Q}[u, v]$ and $p_{1} \in \mathbb{Q}[\sqrt{2}][u, v, w]$ be such that $g(x, y) \notin \mathbb{Q}(x+y, x y, \sqrt{2})$ and the equations of (a) hold. Then there are $\rho \in \mathbb{Q}(x, y), \pi \in \mathbb{Q}[\sqrt{2}][u, v]$ and $\pi_{1} \in \mathbb{Q}[\sqrt{2}][u, v, w]$ such that the equations of (a) hold with $r, p, p_{1}$ replaced by $\rho, \pi, \pi_{1}$. (c) Rationalization Lemma. Let $x, y, r \in \mathbb{R}$ and $F \subset \mathbb{R}$ a field containing $x+y, x y, r^{2}$ but not $r$. If $F(r) \cap \mathbb{Q}(x, y) \not \subset F$, then there is $\rho \in \mathbb{Q}(x, y)$ such that $\rho^{2} \in F$ and $F(\rho)=F(r)$. 4.2. Denote $a_{j}=\sigma_{j}\left(x_{1}, x_{2}, x_{3}\right), j=1,2,3$. (a) Let $x_{1}, x_{2}, x_{3}, r \in \mathbb{R}, p, g \in \mathbb{Q}\left[u_{1}, u_{2}, u_{3}\right]$ and $p_{1} \in \mathbb{Q}\left[u_{1}, u_{2}, u_{3}, v\right]$ be such that $g\left(x_{1}, x_{2}, x_{3}\right) \notin \mathbb{Q}\left(a_{1}, a_{2}, a_{3}\right)$ and $$ \left\{\begin{array}{l} r^{2}=p\left(a_{1}, a_{2}, a_{3}\right) \\ g\left(x_{1}, x_{2}, x_{3}\right)=p_{1}\left(a_{1}, a_{2}, a_{3}, r\right) \end{array}\right. $$ Then $r \in \mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right)$. (b) Rationalization Lemma. Let $x_{1}, x_{2}, x_{3}, r \in \mathbb{R}$ and $F \subset \mathbb{R}$ a field containing $a_{1}, a_{2}, a_{3}, r^{2}$ but not $r$. If $F(r) \cap \mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right) \not \subset F$, then there is $\rho \in \mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right)$ such that $\rho^{2} \in F$ and $F(\rho)=F(r)$. (c) Proposition. If $x_{1}, x_{2}, x_{3} \in \mathbb{R}$ and $x_{1}$ is $\left\{a_{1}, a_{2}, a_{3}\right\}$-expressible by quadratic real radicals, then $x_{1}$ is $\left\{a_{1}, a_{2}, a_{3}\right\}$-expressible by quadratic real radicals so that every radical is in $\mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right)$. 4.3. (a) Let $x, y, r \in \mathbb{C}, p \in \mathbb{Q}[u, v]$ and $p_{1} \in \mathbb{Q}[u, v, w]$ be such that $$ \left\{\begin{array}{l} r^{3}=p(x+y, x y) \\ x=p_{1}(x+y, x y, r) \end{array}\right. $$ (cf. Problem 2.14.d for $k=3$ ). Then $r \in \mathbb{Q}\left[\varepsilon_{3}\right](x, y)$. (b) Same as (a) with $x=p_{1}(x+y, x y, r)$ replaced by $g(x, y)=$ $p_{1}(x+y, x y, r)$ for some $g \in \mathbb{Q}[u, v]$ such that $g(x, y) \notin \mathbb{Q}(x+y, x y)$. (c) Rationalization Lemma. Let $x, y, r \in \mathbb{C}$ and $F \subset \mathbb{C}$ a field containing $x+y, x y, \varepsilon_{3}, r^{3}$ but not $r$. If $F(r) \cap \mathbb{Q}(x, y) \not \subset F$, then there is $\rho \in \mathbb{Q}(x, y)$ such that $\rho^{3} \in F$ and $F(\rho)=F(r)$. (d) Rationalization Lemma. Same as (c) with $x, y$ replaced by $x_{1}, \ldots, x_{n}$ and $x+y, x y$ replaced by $\sigma_{1}\left(x_{1}, \ldots, x_{n}\right), \ldots, \sigma_{n}\left(x_{1}, \ldots, x_{n}\right)$. (e) Rationalization Lemma. Same as (d) with $r^{3}, \rho^{3}$ replaced by $r^{q}, \rho^{q}$ for a prime $q$ and $\varepsilon_{3}$ replaced by $\varepsilon_{q}$. (f) Proposition. If $$ x_{1}, \ldots, x_{n} \in \mathbb{C}, \quad M:=\left\{\sigma_{1}\left(x_{1}, \ldots, x_{n}\right), \ldots, \sigma_{n}\left(x_{1}, \ldots, x_{n}\right)\right\} $$ and $x_{1}$ is $M$-expressible by radicals, then $x_{1}$ is $M$-expressible by radicals so that every radical is in $\bigcup_{q=3}^{\infty} \mathbb{Q}\left[\varepsilon_{q}\right]\left(x_{1}, \ldots, x_{n}\right)$. 4.4. There are numbers $x, y \in \mathbb{R}$ such that if $p \in \mathbb{Q}[u, v]$ and $p(x, y)=0$, then $p=0$. Such numbers are called algebraically independent over $\mathbb{Q}$. 4.5. (а) Докажите достаточность в критерии 2.13 Галуа разрешимости уравнения. (b) Докажите необходимость в критерии 2.13 для $n \leqslant 4$. (c) Сформулируйте и докажите аналог критерия 2.13 для 1радикальности (т.е. для радикальности с одним извлечением корня). (d) Сформулируйте и докажите вещественный аналог критерия 2.13 . 4.6. Пусть $x_{1}, \ldots, x_{n} \in \mathbb{C}$ - все корни многочлена $A \in \mathbb{Q}[t]$ с учетом кратности, $q$ простое, $r \in \mathbb{C}-\mathbb{Q}, \quad r^{q} \in \mathbb{Q}, \quad U \in \mathbb{Q}[\vec{u}], \mathrm{v}$ $U(\vec{x}) \in \mathbb{Q}[r]-\mathbb{Q}$ и id $\in G \subset \Sigma_{n}$. (а) Верно ли, что если $\sum_{\alpha \in G} U\left(\vec{x}_{\alpha}\right) \in \mathbb{Q}$, то $\sum_{\alpha \in G} U\left(\vec{x}_{\alpha \tau}\right) \in \mathbb{Q}$ для любой перестановки $\tau \in \Sigma_{n}$ ? (b) Верно ли, что если $\prod_{\alpha \in G}\left(t-U\left(\vec{x}_{\alpha}\right)\right) \in \mathbb{Q}[t]$, то $\sum_{\alpha \in G}\left(t-U\left(\vec{x}_{\alpha \tau}\right)\right) \in$ $\mathbb{Q}[t]$ для любой перестановки $\tau \in \Sigma_{n}$ ? ## Toward algorithms of solving algebraic equations presented by A. Enne, A. Chilikov, A. Glebov, A. Skopenkov, B. Vukorepa * ## 1 Introduction and statements of results ### 1.1 What is this collection of problems about There are famous Ruffini, Abel and Galois Theorems 2.7, 1.3, 1.4 ${ }^{1}$ on insolvability of algebraic equations in radicals. They are classical results of algebra, which are interesting for the computer science (theory of symbolic computations). All these theorems are formulated below. The main content of this text is exposition of deep algebraic ideas (more precisely, of Galois theory) via simple and beautiful proofs of these theorems (see [ZSS, §27]). It is the more remarkable that for these proofs one only needs the abilities to prove irrationality, to divide polynomials with a remainder, to take the root of a complex number, to multiply permutations, and to solve systems of linear equations. Even those who will not arrive to a complete proof of main results could solve research problems (see [E2, Es, AB, Ko17, Saf] and the references therein).[^10] Before proving the insolvability of algebraic equations we consider a general way for their solution: Lagrange resolvent method. In fact, the main idea of Abel and Galois is the following: if an equation is solvable in radicals at all, then it is solvable by Lagrange method. Lagrange method is used to construct algorithms, e.g. to recognize whether the equation is solvable by radicals. For practical purposes approximative methods of solving equations are more useful than 'radical formulae'. Besides, the equation can be solved using transcendental functions (see Vieta method [ZSS, $\S 4.2]$ and [PSo]; for further development of these ideas see e.g. [Sk10]). However, the problem of 'solvability in radicals' is interesting as a test problem of the modern theories of symbolic computations and computational complexity. On the novelty. Proofs which are provided in solutions are not assumed to be new (but the reader could find new proofs). However, this text contains many pedagogical inventions (see [ZSS, §5.2.1, 5.2.2]). The proofs are different from proofs which are presented and quoted in [ZSS, §5]. Unfortunately, the proofs presented here are not well-known. Standard textbooks of algebra first expose Galois theory and then use its results to prove these theorems. However, it is much more economic and clearer not only to solve directly quadratic and cubic equations but also to prove corresponding theorems directly ${ }^{2}$. Of course, for such direct proofs, one should re-discover and use key ideas of Galois theory.[^11] ### 1.2 Insolvability in real radicals A real number is called expressible by real radicals if it can be obtained using number 1 and operations of addition, subtraction, multiplication, division by a non-zero number, and taking the $n$-th root of a positive number, where $n$ is a positive integer. In other words, a real number $a$ is expressible by real radicals if some set containing this number can be obtained starting from the set $\{1\}$ and using the following operations. To a given set $M \subset \mathbb{R}$ containing numbers $x, y \in M$ one can add $$ \begin{aligned} & \text { numbers } \quad x+y, x-y, x y, \quad \text { number } x / y \text { when } y \neq 0 \\ & \text { and number } \sqrt[n]{x} \text { for } x>0 \text { and integer } n>0 \end{aligned} $$ A number $a$ is expressible by real radicals if and only if there exist - positive integers $s, k_{1}, \ldots, k_{s}$, - real numbers $f_{1}, \ldots, f_{s}$ and polynomials $p_{0}, p_{1}, \ldots, p_{s}$ with rational coefficients of $0,1, \ldots, s$ variables respectively such that $$ \left\{\begin{array}{l} f_{1}^{k_{1}}=p_{0} \\ f_{2}^{k_{2}}=p_{1}\left(f_{1}\right) \\ \cdots \\ f_{s}^{k_{s}}=p_{s-1}\left(f_{1}, \ldots, f_{s-1}\right) \\ a=p_{s}\left(f_{1}, \ldots, f_{s}\right) \end{array}\right. $$ Remark 1.1. (a) Any real root of a quadratic equation with rational coefficients is expressible by real radicals. (b) The equation $x^{3}+x+1=0$ has exactly one real root which is expressible by real radicals [ZSS, §4.2], see also Problem 2.8 (c). (c) The equation $x^{4}+4 x-1=0$ has two real roots; both of them are expressible by real radicals [ZSS, §4.2], see also Problem 2.10 (d). (d) Any real constructible number $[\mathrm{ZSS}, \S 5.1 .2]$ is expressible by real radicals. (e) There exists a cubic polynomial with rational coefficients such that none of its roots is expressible by real radicals (for example, $x^{3}-$ $3 x+1$ ). (This statement is proven in Remark (f).) (f) The number $\cos (2 \pi / 9)$ is not expressible by real radicals. Let us apply the triple-angle formula for cosine. Then the numbers $\cos (2 \pi / 9), \cos (8 \pi / 9), \cos (14 \pi / 9)$ are the roots of the polynomial $8 y^{3}-$ $6 y+1=0$. By Theorem 1.2 none of these numbers is expressible by real radicals. (g) The trisection of an angle is impossible in real radicals. That is, there exists a number $\alpha$ (for example, $\alpha=2 \pi / 3$ ) such that the number $\cos \alpha$ is expressible by real radicals and the number $\cos (\alpha / 3)$ is not expressible by real radicals. (This statement follows from Remark (f).) Theorem 1.2 (solvability in real radicals). For a cubic polynomial with rational coefficients the following conditions are equivalent: (i) the polynomial has either at least one rational root or exactly one real root; (ii) the polynomial has a root which is expressible by real radicals; (iii) all the real roots of the polynomial are expressible by real radicals. The uniqueness of the real root of the 'shortened' equation $x^{3}+$ $p x+q=0$ is equivalent to the following condition: ' $p=q=0$ or $(p / 3)^{3}+(q / 2)^{2}>0^{\prime}$ [ZSS, Problem 8.1.5.d]. Obviously, $(i i) \Leftrightarrow(i i i)$. This follows from Remark 1.1.a. The solvability in Theorem 1.2 (that is $(i) \Rightarrow(i i)$ ) can be proved by del Ferro method [ZSS, §4.2]; see another proof in $\S 2.3$. The insolvability in Theorem 1.2 (that is $(i i) \Rightarrow(i))$ has more complicated proof. It is easier to prove the similar result on insolvability in polynomials, see $\S 2.5$. ### 1.3 Insolvability in complex radicals Now consider formulae which involve complex numbers. It turns out that a cubic equation (for example, $x^{3}-3 x+1$ ) that is not solvable in real radicals can be solved in complex radicals. A complex number is called expressible by radicals if it can be obtained using number 1 and operations of addition, subtraction, multiplication, division by a non-zero number and taking the $n$-th root, where $n$ is a positive integer. In other words, a complex number $a$ is expressible by radicals if some set containing this number can be obtained starting from the set $\{1\}$ and using the following operations. To a given set $M \subset \mathbb{C}$ containing numbers $x, y \in M$ one can add numbers $\quad x+y, x-y, x y, \quad$ number $x / y$ when $y \neq 0$, and any number $r \in \mathbb{C}$ such that $r^{n}=x$ for some integer $n>0$. For example, any (complex) root of a quadratic equation with rational coefficients is expressible by radicals. Similar assertions hold for equations of 3-rd and 4-th degree. These assertions can be proved by del Ferro and Ferrari methods [ZSS, §4.2]; see another proof in $\S 2.3$. However, analogous assertions for equations of higher degrees do not hold. Theorem 1.3 (Galois). There exists an equation of 5-th degree with rational coefficients (for example, $x^{5}-4 x+2=0$ ) neither of whose roots are expressible by radicals. The famous problem of solvability in radicals was solved by weaker Ruffini-Abel theorems proved a little earlier. The Ruffini Theorem 2.7 has more complicated statement. But it leads us to the proof of Galois theorem. The precise statement of Abel theorem is even more complicated. It is not presented here, see [Sk15, Remark 7]. An easier way to solve the solvability problem is to prove the following Galois Theorem 1.4. This theorem is weaker than Galois Theorem 1.3 and has an easier proof. For $X \subset \mathbb{C}$, a complex number $a$ is called $X$ expressible by radicals if $a$ can be expressed using the set $X \cup\{1\}$ and the operations from the definition of the expressibility by radicals. Theorem 1.4 (Galois). There are $a_{0}, a_{1}, a_{2}, a_{3}, a_{4} \in \mathbb{C}$ such that no root of the equation $x^{5}+a_{4} x^{4}+\ldots+a_{1} x+a_{0}=0$ is $\left\{a_{0}, a_{1}, \ldots, a_{4}\right\}$ expressible by radicals. Theorem 1.5. There is an algorithm to recognize whether all the roots of the equation $x^{n}+a_{n-1} x^{n-1}+\ldots+a_{1} x+a_{0}=0$ are expressible by radicals (if $a_{n-1}, \ldots, a_{0} \in \mathbb{Q}$ are known). Theorem 1.5 is proved using Galois Solvability Criterion 2.13.b and an estimation of the number of operations. ### 1.4 Plan This project consists of three formally independent parts. In the first two parts the definition of 'expressibility by radicals' from $\S 2.2$ is used. (1) In $\S 2.3$ we discuss Lagrange's resolvent method used to solve equations. Formally this method is not used to prove insolvability. However, familiarity with this method is useful because - proofs of insolvability were invented during the analysis of this method, - this familiarity helps to check intermediate conjectures (which occur in attempts to prove insolvability), and - this method is required to prove Theorem 1.5. (2) The proof of the Ruffini theorem 2.7 is based on the idea of symmetry and is sketched in $\S 3.1$. This proof is prepared by $\S 2.5$. Subsection 2.4 prepares to $\S 2.5$ and to the proof of Galois Theorem 1.4 . (3) The proof of Theorem 1.2 on solvability in real radicals is based on the idea of conjugation (or of algebraic symmetry). This proof is prepared by $\S 2.1, \S 3.2$ and $\S 3.3$. Theorems 1.3 and 1.5 are not proved here. See the proof of Theorem 1.3 in [ZSS, §5], [Sk19, §9]. Galois Theorem 1.4 is proved in additional problems, cf. [Sk15], [Sk19, §5]. The proof is based on reduction to Ruffini Theorem 2.7 using the idea of conjugation ( $\S 2.1, \S 3.2$ and §3.3). ### 1.5 Recommendations for participants For every solution which has been written down and marked with either ' + ' or ' + .' a student (or a group of students) get a 'bean'. The jury may also award extra bean for beautiful solutions, solutions of hard problems, or solutions typeset in $\mathrm{T}_{\mathrm{E}} \mathrm{X}$. The jury has infinitely many beans. One may submit a solution in oral form, but one loses a bean with each 5 attempts (successful or not). If a problem is marked by bold and named 'theorem' ('lemma', 'corollary', etc.), then this statement is important. Usually we provide (as a problem) the formulation of beautiful or important statement before its proof. In this case to prove this statement one possibly needs to solve next problems. If you are stuck on a certain problem, try looking at the next ones. They may turn out to be helpful. We suggest to all the students working on the project to consult the jury on any questions on the project. Students who successfully work on the project will get interesting extra problems. Please notify us if you already know solutions of several problems. If you confirm your knowledge by presenting some of them, you will be allowed not to receive plus-marks for their solutions, but to use them in solutions of other problems. ## 2 Problems before the semifinal In this text equality signs involving polynomial $f$ (or $f_{j}$ ) mean equality of polynomials (i.e. componentwise equality). In $\$ \S 2.1,3.2$ and 3.3 'polynomial with rational coefficients' is called a 'polynomial'. Denote $$ \varepsilon_{q}:=\cos (2 \pi / q)+i \sin (2 \pi / q) $$ ### 2.1 Representability using only one square root 2.1. Can the following number be represented as $a+\sqrt{b}$ with $a, b \in$ $\mathbb{Q}:$ (a) $\sqrt{3+2 \sqrt{2}}$; (b) $\frac{1}{7+5 \sqrt{2}}$; (c) $\sqrt[3]{7+5 \sqrt{2}}$; (d) $\cos (2 \pi / 5)$; (e) $\sqrt[3]{2}$ (f) $\sqrt{2}+\sqrt[3]{2}$ (g) $\cos (2 \pi / 9)$; (h) $\sqrt{2+\sqrt{2}}$; (i) $* \cos (2 \pi / 7)$; (j) $\sqrt{2}+\sqrt{3}+\sqrt{5}$ ? Lemma 2.2. Assume that $r \in \mathbb{R}-\mathbb{Q}$ and $r^{2} \in \mathbb{Q}$. (a) Irreducibility. The polynomial $x^{2}-r^{2}$ is irreducible over $\mathbb{Q}$. (b) Linear independence. If $a, b \in \mathbb{Q}$ and $a+b r=0$, then $a=b=0$. (c) If $r$ is a root of a polynomial, then this polynomial is divisible by $x^{2}-r^{2}$. (d) Conjugation. If $r$ is a root of a polynomial, then $-r$ is also its root. (e) Conjugation. If $a, b \in \mathbb{Q}$ and a polynomial has a root $a+b r$, then $a-b r$ is also a root of this polynomial. (f) If $a, b \in \mathbb{Q}$ and a cubic polynomial has a root $a+b r$, then this polynomial has a rational root. Theorem 2.3. If a polynomial of degree at least 3 is irreducible over $\mathbb{Q}$, then none of its roots equals to $a \pm \sqrt{b}$ for some $a, b \in \mathbb{Q}$. Lemma 2.4 (Extension). Suppose we can obtain a number using number 1, several operations of addition, subtraction, multiplication, division by a non-zero number and exactly one operation of taking the square root of a positive number. Then the number can be represented as $a \pm \sqrt{b}$, where $a, b \in \mathbb{Q}$ and $b>0$. 2.5.* Find all $n$ such that the number $\cos (2 \pi / n)$ can be represented as $a+\sqrt{b}$, where $a, b \in \mathbb{Q}$. ## Hints to $\S 2.1$ 2.2. (a) If the polynomial $x^{2}-r^{2}$ is reducible over $\mathbb{Q}$, then it has a rational root. This is a contradiction. (b) If $b \neq 0$, then $r=-a / b \in \mathbb{Q}$, which is impossible. Hence $b=0$, thus $a=0$. (c) Divide our polynomial with a remainder ${ }^{3}$ by $x^{2}-r^{2}$ : $$ P(x)=\left(x^{2}-r^{2}\right) Q(x)+m x+n $$ Substitute $x=r$. By the Linear independence lemma (see (b)) the reminder is equal to zero. (d) By (c) if $R^{2}=r^{2}$, then $R$ is a root of the polynomial. (e) Let $P$ be given polynomial, and set $G(t):=P(a+b t)$. Then $G(r)=0$. Hence by (d) we obtain $G(-r)=0$. (f) If $b=0$ the assertion is proved. Otherwise by (e) the polynomial has the roots $a \pm b r$. These roots are distinct. Hence the third root is rational by the Vieta theorem. 2.4. It would suffice to prove that the set of all numbers of the form $a \pm \sqrt{b}$ is closed under operations of addition, subtraction, multiplication and division. This is obviously false: $(1+\sqrt{2})+(1+\sqrt{3})$ cannot be represented as $a \pm \sqrt{b}$, where $a, b \in \mathbb{Q}$ (prove this!). ### 2.2 Definition of the expressibility by radicals for polynomials The solution of quadratic equation can be expressed by the following formulae: $$ (x-y)^{2}=(x+y)^{2}-4 x y \quad \text { and } \quad x=\frac{x+y+(x-y)}{2} $$[^12] These formulae show that the root $x$ of a quadratic equation is expressible by radicals using the coefficients $x+y, x y$ of the equation. The rigorous definition of expressibility by radicals is given below. Denote the elementary symmetric polynomials by $$ \sigma_{1}\left(x_{1}, \ldots, x_{n}\right):=x_{1}+\ldots+x_{n}, \quad \ldots, \quad \sigma_{n}\left(x_{1}, \ldots, x_{n}\right)=x_{1} \cdot \ldots \cdot x_{n} $$ If the number $n$ and the arguments $x_{1}, \ldots, x_{n}$ are clear from the context, we omit them from the notation. A polynomial $p \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ is called expressible by (complex) radicals, if one can add $p$ to the collection $\left\{\sigma_{1}, \ldots, \sigma_{n}\right\} \cup \mathbb{C}$ of polynomials by a sequence of the following operations: - if the polynomials $f$ and $g$ are already contained in the collection one can add their sum $f+g$ and their product $f g$; - if the polynomial $g$ is already contained in the collection and $g=f^{k}$ for some $f \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ and integer $k>1$ one can add $f$ to the collection. Remark 2.6. (a) E.g. if a collection contains $x^{2}+2 y$ and $x-y^{3}$, then one may apply the first operations and add the polynomial $$ -5\left(x^{2}+2 y\right)^{2}+3\left(x^{2}+2 y\right)\left(x-y^{3}\right)^{6} $$ to the collection; moreover, if a collection already contains $x^{2}-2 x y+y^{2}$, then one may apply the second operation and add $x-y$ (or $y-x)$. (b) If we use only first operations we can add a polynomial with complex coefficients of polynomials which are already available. (c) By Vieta theorem $\sigma_{1}, \ldots, \sigma_{n}$ are the coefficients of the polynomial $$ t^{n}-\sigma_{1} t^{n-1}+\ldots+(-1)^{n-1} \sigma_{n-1} t+(-1)^{n} \sigma_{n} \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right][t] $$ with roots $x_{1}, \ldots, x_{n}$. Therefore, the expressibility by radicals of the polynomial $x_{1}$ is equivalent to the expressibility (in the above sense) of its root $x_{1}$ in terms of the coefficients of this polynomial. (d) The polynomial $x_{1}$ is expressible by radicals if and only if there exist: - positive integers $s, k_{1}, \ldots, k_{s}$, - polynomials $f_{1}, \ldots, f_{s}$ and $p_{0}, p_{1}, \ldots, p_{s}$ with complex coefficients of $n$ and of $n, n+1, \ldots, n+s$ variables respectively, such that $$ \left\{\begin{array}{l} f_{1}^{k_{1}}=p_{0}\left(\sigma_{1}, \ldots, \sigma_{n}\right) \\ f_{2}^{k_{2}}=p_{1}\left(\sigma_{1}, \ldots, \sigma_{n}, f_{1}\right) \\ \cdots \\ f_{s}^{k_{s}}=p_{s-1}\left(\sigma_{1}, \ldots, \sigma_{n}, f_{1}, \ldots, f_{s-1}\right) \\ x_{1}=p_{s}\left(\sigma_{1}, \ldots, \sigma_{n}, f_{1}, \ldots, f_{s}\right) \end{array}\right. $$ Here we omit the variables $\left(x_{1}, \ldots, x_{n}\right)$ of the polynomials $\sigma_{1}, \ldots, \sigma_{n}, f_{1}, \ldots, f_{s}$. (e) Given $x+y$ and $x y$, is it always possible to find $x$ ? A simple formalization of this question is the following: does there exist a map $f: \mathbb{R}^{2} \rightarrow \mathbb{R}$ such that $f(x+y, x y)=x$ for any $x, y \in \mathbb{R}$ ? The answer is no. Indeed, consider the pairs $x=1, y=2$ and $x=$ $2, y=1$. Therefore, the expressibility by radicals does not allow 'to find $\mathrm{x}^{\prime}$ in the sense described above. Analogously, given $\sigma_{1}=x+y+z, \sigma_{2}=x y+y z+z x$ and $\sigma_{3}=x y z$ is it not always possible to find $(x-y)(y-z)(z-x)$. Indeed, consider the triples $x=0, y=1, z=-1$ and $x=0, y=-1, z=1$. Theorem 2.7 (Ruffini). For every positive integer $n \geqslant 5$ the polynomial $x_{1}$ is not expressible by radicals. The proof shows that even the polynomial $x_{1} x_{2}+x_{2} x_{3}+x_{3} x_{4}+$ $x_{4} x_{5}+x_{5} x_{1}$ is not expressible by radicals for $n=5$. ### 2.3 Solution of equations of low degrees 2.8. Which of the following polynomials are expressible by radicals for $n=3$ ? (a) $(x-y)(y-z)(z-x)$; (b) $x^{9} y+y^{9} z+z^{9} x$; (c) $x$. To solve Problem 2.8 and the following problems, one can use the fundamental theorem on symmetric polynomials,see for example [Sk19, 4.6.3c]. Hints for part (c) are Problems 2.9.a and 2.11.c. 2.9. A polynomial $f \in \mathbb{C}\left[u_{1}, u_{2}, \ldots, u_{n}\right]$ is called cyclic symmetric if $f\left(u_{1}, u_{2}, \ldots, u_{n}\right)=f\left(u_{2}, u_{3}, \ldots, u_{n-1}, u_{n}, u_{1}\right)$. (a) Find at least one pair $\alpha, \beta \in \mathbb{C}$ such that the polynomial $(u+$ $v \alpha+w \beta)^{3}$ is cyclic symmetric, but the polynomial $u+v \alpha+w \beta$ is not. (b) Express $x_{1} x_{3}+x_{3} x_{5}+x_{5} x_{7}+x_{7} x_{9}+x_{9} x_{1}$ by operations from the definition of expressibility by radicals starting with some cyclic symmetric polynomials in $x_{1}, x_{2}, \ldots, x_{10}$. 2.10. Which of the following polynomials are expressible by radicals for $n=4$ ? (a) $(x-y)(x-z)(x-t)(y-z)(y-t)(z-t)$; (b) $x y+z t$; (c) $x+y-z-t$; (d) $x$. 2.11. Solve the following systems of equations $(x, y, z, t$ are unknowns, $a, b, c, d$ are known): $$ \begin{gathered} (\mathrm{a})\left\{\begin{array} { l } { x + y + z + t = a , } \\ { x + y - z - t = b , } \\ { x - y + z - t = c , } \\ { x - y - z + t = d } \end{array} \quad \quad \quad \quad \left\{\begin{array}{l} x+y+z+t=a \\ x+i y-z-i t=b \\ x-y+z-t=c \\ x-i y-z+i t=d \end{array}\right.\right. \\ \text { (c) }\left\{\begin{array}{l} x+y+z=a \\ x+\varepsilon_{3} y+\varepsilon_{3}^{2} z=b \\ x+\varepsilon_{3}^{2} y+\varepsilon_{3} z=c \end{array}\right. \end{gathered} $$ Expressions from Problem 2.11 are called Lagrange resolvents. They are 'better' than roots because they are 'more symmetric' in the following sense. Solution of cubic equation using Lagrange resolvents (solution of Problem 2.8(c)). To find the roots $x, y, z$ of a cubic equation, it suffices to find the expressions $a, b, c$ from Problem 2.11 (c). Notice that the del Ferro method from [Sk19, 4.2.2] leads us to the same expressions. By Vieta theorem, $a=a(x, y, z)$ is the coefficient of the equation. Under the substitution $x \leftrightarrow y$, polynomial $b=b(x, y, z)$ goes to $\varepsilon_{3} c$, and $c=c(x, y, z)$ goes to $\varepsilon_{3}^{2} b$ (check this!). Therefore, the polynomials $b c$ and $b^{3}+c^{3}$ do not change under this substitution. Analogously, they do not change under substitution $z \leftrightarrow y$. Therefore the polynomials $b c$ and $b^{3}+c^{3}$ are symmetric, i.e., they do not change under any permutation of variables. From the theorem on representability of a symmetric polynomial as a polynomial in elementary symmetric polynomials (see e.g. [Sk19, 4.6.3c]) and Vieta theorem it follows that the $b c$ and $b^{3}+c^{3}$ polynomials in $x, y, z$ can be represented as polynomials in the coefficients of the equation. Hence we can obtain $b^{3}$ and $c^{3}$ by solving certain quadratic equation. Now, by solving certain quadratic equation we can obtain $b^{3}$ and $c^{3}$. By Ruffini Theorem 2.7, Lagrange resolvent method demonstrated by solving equations of degrees 3 and 4 (Problems 2.8 (c) and 2.10 (d)) does not work for degree 5. Guess why! Denote by $\Sigma_{q}$ the set of permutations of the set $\{1,2, \ldots, q\}$. For a permutation $\alpha \in \Sigma_{q}$ denote $$ \vec{u}_{\alpha}:=\left(u_{\alpha(1)}, \ldots, u_{\alpha(q)}\right) $$ Define the Lagrange resolvent by $$ t\left(u_{1}, \ldots, u_{q}\right):=\varepsilon_{q} u_{1}+\varepsilon_{q}^{2} u_{2}+\ldots+\varepsilon_{q}^{q} u_{q} $$ Define Galois resolvent by $$ Q\left(u_{1}, \ldots, u_{q}, y\right):=\prod_{\alpha \in \Sigma_{q}}\left(y-t\left(\vec{u}_{\alpha}\right)\right) \in \mathbb{Q}\left[\varepsilon_{q}\right]\left[u_{1}, \ldots, u_{q}, y\right] $$ 2.12. (a) We have $Q\left(\varepsilon_{q} u_{1}, \ldots, \varepsilon_{q} u_{q}, y\right)=Q\left(u_{1}, \ldots, u_{q}, y\right)$. (b) For some $R_{Q} \in \mathbb{Q}\left[\varepsilon_{q}\right]\left[u_{1}, \ldots, u_{q}, z\right]$ we have $Q\left(u_{1}, \ldots, u_{q}, y\right)=$ $R_{Q}\left(u_{1}, \ldots, u_{q}, y^{q}\right)$. (c) If $x_{1}, \ldots, x_{q} \in \mathbb{C}$ are the roots of a polynomial $f \in \mathbb{Q}[x]$ of degree $q$, then $Q\left(x_{1}, \ldots, x_{q}, y\right) \in \mathbb{Q}\left[\varepsilon_{q}\right][y]$ and even $Q\left(x_{1}, \ldots, x_{q}, y\right) \in \mathbb{Q}[y]$. The polynomial $R_{Q}\left(x_{1}, \ldots, x_{q}, z\right) \in \mathbb{Q}[z]$ is called the resolvent polynomial for $f$. (d)* All the roots of the resolvent polynomial for $f(x)=x^{5}+15 x+$ 11 (and therefore, all the roots of $f$ ) are expressible by radicals. Theorem 2.13.* (a) For $a, b \in \mathbb{R}$ all the roots of the equation $x^{5}+a x+b=0$ are expressible by radicals if and only if $a=\frac{15 \pm 20 c}{c^{2}+1}$ and $b=\frac{44 \mp 8 c}{c^{2}+1}$ for some $c \in \mathbb{Q}, c \geqslant 0$. (b) (Galois Solvability Criterion) For each $a_{n-1}, \ldots, a_{0} \in \mathbb{Q}$ all the roots of the equation $A(x):=x^{n}+a_{n-1} x^{n-1}+\ldots+a_{1} x+a_{0}=0$ are expressible by radicals if and only if a set of degree 1 polynomials over $\mathbb{Q}$ can be obtained from $\{A\}$ using the following operations: - (factorization) if one of our polynomials equals to $P_{1} P_{2}$ for some non-constant $P_{1}, P_{2} \in \mathbb{Q}[x]$, then replace $P_{1} P_{2}$ by $P_{1}$ and $P_{2}$; - (extracting a root) if one of our polynomials equals to $P\left(x^{q}\right)$ for some $P \in \mathbb{Q}[x]$, then replace $P\left(x^{q}\right)$ by $P(x)$; - (taking Galois resolvent) replace one of our polynomials $P$ by the polynomial $Q\left(y_{1}, \ldots, y_{q}, y\right)$, where $y_{1}, \ldots, y_{q}$ are all the roots of $P$. (Analogously to Problem 2.12.c $Q\left(y_{1}, \ldots, y_{q}, y\right) \in \mathbb{Q}[y]$.) Part (a) is derived from (b) [PSo]. Part 'if' in (b) is easier. This part can be proved using Lagrange resolvent method which is considered in this subsection. Part 'only if' in (b) is more complicated. This can be proved similarly to Galois Theorems 1.3, 1.4. ### 2.4 There is only one way to solve quadratic equation Systems of equations studied here and in the following subsection arise when solving equations by radicals ('using one radical'), see Remark 2.6.d. 2.14. (a,b) Solve the system of equations in polynomials $f(x, y)$, $p(u, v)$ and $q(u, v, w)$ with real coefficients: (a) $\left\{\begin{array}{l}f^{2}(x, y)=p(x+y, x y) \\ x=q(x+y, x y, f(x, y))\end{array}\right.$. (b) $\left\{\begin{array}{l}f^{k}(x, y)=p(x+y, x y) \\ x=q(x+y, x y, f(x, y))\end{array}\right.$, where $k>0$ is an integer. $\left(\mathrm{c}, \mathrm{d}^{*}\right)$ Solve the analogues of $(\mathrm{a}, \mathrm{b})$ where the polynomial $f$ is replaced by a function $f: \mathbb{R}^{2} \rightarrow \mathbb{R}$ (which is not assumed to be continuous). The system of equations from 2.14.a is satisfied, for example, by the polynomials $$ f(x, y)=x-y, \quad p(u, v)=u^{2}-4 v \quad \text { and } \quad q(u, v, w)=\frac{u+w}{2} $$ 2.15. Assume that $f, g \in \mathbb{R}[x, y]$. (a) Lemma. If $f g=0$, then $f=0$ or $g=0$. Warnings. There exist functions $F, G: \mathbb{R} \rightarrow \mathbb{R}$ such that $F G=$ $0, F \neq 0$ and $G \neq 0$. There exist two different polynomials in two variables which are equal at an infinite set of points. Do not use without proof the fact that if the values of polynomials in two variables are equal in any point, then the polynomials are equal. (b) If $f^{2}=g^{2}$, then $f=g$ or $f=-g$. (c) If $f^{2}+f g+g^{2}=0$, then $f=0$ or $g=0$. (d) If $f^{3}=g^{3}$, then $f=g$. (e) If $f^{5}=g^{5}$, then $f=g$. (f) $f^{5}-g^{5}=(f-g)\left(f-\varepsilon_{5} g\right)\left(f-\varepsilon_{5}^{2} g\right)\left(f-\varepsilon_{5}^{3} g\right)\left(f-\varepsilon_{5}^{4} g\right)$. To prove the assertions 2.14.bd, the following notions and lemma are useful. A polynomial $f$ in two variables $x, y$ is called symmetric, if $f(x, y)=$ $f(y, x)$. A polynomial $f$ is called antisymmetric, if $f(x, y)=-f(y, x)$. 2.16. (a) Lemma. If $f \in \mathbb{R}[x, y]$ is a polynomial with real coefficients in two variables such that $f^{2}$ is symmetric, then $f$ is either symmetric or antisymmetric. (b) Lemma. If $f \in \mathbb{R}[x, y]$ is such that $f^{2 k+1}$ is symmetric, then $f$ is symmetric. (c) If $f \in \mathbb{R}[x, y]$ is antisymmetric, then there exists a symmetric polynomial $a \in \mathbb{R}[x, y]$ such that $f=(x-y) a$. To prove the assertions above and to solve other problems, Lemma 2.15. a would be useful. 2.17. For which of the statements 2.15 and 2.16 their analogues for polynomials with complex coefficients hold? Now we give a generalized form of assertion 2.14 for an arbitrary number of steps in the definition of the expressibility by radicals. 2.18. A rational fraction is a 'formal ratio of polynomials', i.e. a pair $f / g:=(f, g)$ of polynomials with $g \neq 0$. We define $f / g \sim f^{\prime} / g^{\prime}$ if and only if $f g^{\prime}=f^{\prime} g$. The polynomial $f$ is identified with the pair $(f, 1)$. Denote by $\mathbb{R}\left(u_{1}, \ldots, u_{n}\right)$ the set of all rational fractions with real coefficients in variables $u_{1}, \ldots, u_{n}$. (a) Define the sum and the product of rational fractions. Are they well-defined? Check this! (b) Consider the system of Remark 2.6.(d) for $n=2$, where $f_{j}$ and $p_{j}$ are rational functions (not necessarily polynomials). Assume that the system is minimal. This means that there is no system with a smaller $s$, and that $f_{j}^{k}$ is not a rational fraction of $x+y, x y, f_{1}, \ldots, f_{j-1}$ for any $j=1, \ldots, s$ and $k0$. For the proof the following definition and statement are useful. A polynomial $f \in \mathbb{R}[x, y, z]$ is called cyclic symmetric if $f(x, y, z)=$ $f(y, z, x)$. 2.20. If $f \in \mathbb{R}[x, y, z]$ and the polynomial (a) $f^{3}$; (b) $f^{2}$ is cyclic symmetric, then $f$ is cyclic symmetric. Remark 2.21 (cf. solution of Problem 2.8.c). There are no polynomials $f_{1}(x, y, z), \quad f_{2}(x, y, z), \quad p_{0}(u, v, w), \quad p_{1}\left(u, v, w, \tau_{1}\right), \quad p_{2}\left(u, v, w, \tau_{1}, \tau_{2}\right)$ with real coefficients such that $$ \left\{\begin{array}{l} f_{1}^{2}=p_{0}\left(\sigma_{1}, \sigma_{2}, \sigma_{3}\right) \\ f_{2}^{3}=p_{1}\left(\sigma_{1}, \sigma_{2}, \sigma_{3}, f_{1}\right) \\ x=p_{2}\left(\sigma_{1}, \sigma_{2}, \sigma_{3}, f_{1}, f_{2}\right) \end{array}\right. $$ A generalization of Remark 2.21 to an arbitrary number of steps can be formalized by the definition of expressibility by real radicalswhich is obtained from its complex analogue ( $\$ 2.3$ ) by replacing complex coefficients by real coefficients. The formulae at the beginning of $\S 2.2$ show that $x$ is expressible by real radicals for $n=2$. The solution of Problem 2.8.ab shows that both polynomials $$ (x-y)(y-z)(z-x) \quad \text { and } \quad x^{9} y+y^{9} z+z^{9} x $$ are expressible by real radicals for $n=3$. Theorem 2.22. The polynomial $x$ is not expressible by real radicals for $n=3$. Theorem 2.22 is yet another formalization of the fact that a root of a cubic equation is not expressible by real radicals via its coefficients, see Remark 1.1.e. Theorem 2.22 is implied by the following lemma. Lemma 2.23 (keeping cyclic symmetry). If $q>0$ is an integer, $f \in \mathbb{R}[x, y, z]$ and the polynomial $f^{q}$ is cyclic symmetric, then $f$ is cyclic symmetric. 2.24. For which of the statements from this subsection their analogues for polynomials with complex coefficients hold? ## Toward algorithms of solving algebraic equations presented by A. Enne, A. Skopenkov, A. Glebov, A. Chilikov, B. Vukorepa ## 3 Problems after the semifinal ### 3.1 Insolvability 'in polynomials' The definition of expressibility by radicals for a polynomial is given in §2.2. Formally, the Ruffini Theorem 2.7 follows from Lemma 3.4. The most difficult and interesting task is to invent the statement of this lemma. In order to do that we prove the following simple facts. Explain why the polynomial $x$ is not a polynomial of $x+y$ and $x y$. 3.1. The polynomial $x_{1}$ is not expressible by radicals in such a way that the second operation in the definition of expressibility is applied only for (a) $k=2$ (hint: see Problem 2.24); (b) $k=3$. 3.2. Which of the following assertions are true for every $f \in \mathbb{C}\left[x_{1}, \ldots, x_{5}\right]$ ? (a) If $f^{3}$ is cyclic symmetric, then $f$ is cyclic symmetric. (b) If $f^{5}$ is cyclic symmetric, then $f$ is cyclic symmetric. (c) If $f^{3}$ is symmetric, then $f$ is symmetric. (d) If $f^{2}$ is symmetric, then $f$ is symmetric. A cycle of length 3 is a permutation of an $n$-element set which moves some 3 elements cyclically and does not change positions of any other elements. A polynomial $f \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ is even symmetric if for any cycle $\alpha$ of length 3 the polynomials $f\left(x_{1}, x_{2}, \ldots, x_{n}\right)$ and $f\left(x_{\alpha(1)}, x_{\alpha(2)}, \ldots, x_{\alpha(n)}\right)$ are equal. 3.3. (a) Find a cyclic symmetric polynomial that is not even symmetric. (b) If a permutation does not change the polynomial from your solution of Problem 3.2.d, then it can be represented as a composition of cycles of length 3 . Lemma 3.4 (keeping even symmetry). If $q>0$ is an integer, $f \in$ $\mathbb{C}\left[x_{1}, \ldots, x_{5}\right]$, and the polynomial $f^{q}$ is even symmetric, then $f$ is even symmetric. 3.5. Suppose $f \in \mathbb{C}\left[x_{1}, \ldots, x_{n}\right]$ is a polynomial. (a) If the polynomial $f^{7}$ is even symmetric, then $f$ is even symmetric. (b) If $n \geqslant 5$ and the polynomial $f^{3}$ is even symmetric, then $f$ is even symmetric. (c) If $n \geqslant 5$, then any cycle of length 3 on an $n$-element set can be written as a product of permutations of the form $(a b)(c d)$, where $a, b, c, d$ are pairwise distinct (i.e. as a product of compositions of transpositions with disjoint supports). 3.6. The definition of rational expressibility by real (complex) radicals is analogous to the definition of expressibility by radicals. Polynomials are replaced by rational fractions (with appropriate coefficients; see the definition in Problem 2.18). Is the polynomial $x_{1}$ rationally expressible by (a) real radicals for $n=3$ ? (b) (complex) radicals for $n=5$ ? ### 3.2 Representability using only one cubic root Here we develop the ideas from $\S 2.1$. 3.7. Can the following number be represented as $a+b \sqrt[3]{2}+c \sqrt[3]{4}$ with $a, b, c \in \mathbb{Q}$ : (a) $\sqrt{3}$; (b) $\frac{1}{1+5 \sqrt[3]{2}+\sqrt[3]{4}}$; (c) $\cos (2 \pi / 9)$; (d) $\sqrt[5]{3}$; (e) $\sqrt[3]{3}$; (f) the maximal real root of $x^{3}-4 x+2=0$; (g)* the unique real root of $x^{3}-6 x-6=0$; (h) ${ }^{*}$ the unique real root of $x^{3}-9 x-12=0$ ? Lemma 3.8. Assume that $r \in \mathbb{R}-\mathbb{Q}$ and $r^{3} \in \mathbb{Q}$. (a) Irreducibility. The polynomial $x^{3}-r^{3}$ is irreducible over $\mathbb{Q}$. (b) Linear independence. If $a+b r+c r^{2}=0$ with $a, b, c \in \mathbb{Q}$, then $a=b=c=0$. $\left(\mathrm{b}^{\prime}\right)$ Linear independence over $\mathbb{Q}\left[\varepsilon_{3}\right]$. If $$ k, \ell, m \in \mathbb{Q}\left[\varepsilon_{3}\right]:=\left\{u+v \varepsilon_{3}: u, v \in \mathbb{Q}\right\} $$ and $k+\ell r+m r^{2}=0$, then $k=\ell=m=0$. (c) If $r$ is a root of a polynomial, then this polynomial is divisible by $x^{3}-r^{3}$. (d) Conjugation. If $r$ is a root of a polynomial, then the numbers $\varepsilon_{3} r$ and $\varepsilon_{3}^{2} r$ are also its roots. (e) Conjugation. If $a, b, c \in \mathbb{Q}$ and a polynomial has root $x_{0}:=$ $a+b r+c r^{2}$, then the numbers $$ x_{1}:=a+b \varepsilon_{3} r+c \varepsilon_{3}^{2} r^{2} \quad \text { and } \quad x_{2}:=a+b \varepsilon_{3}^{2} r+c \varepsilon_{3} r^{2} $$ are also its roots. (f) Rationality. If $a, b, c \in \mathbb{Q}$, then the number $a+b r+c r^{2}$ is a root of some cubic polynomial. Theorem 3.9. If a polynomial is irreducible over $\mathbb{Q}$ and has a root $a+b r+c r^{2}$ for some $r \in \mathbb{R}-\mathbb{Q}$ and $a, b, c, r^{3} \in \mathbb{Q}$, then this polynomial is cubic and it has exactly one real root. Lemma 3.10 (Extension). A number expressible by real radicals with only one extraction of a cubic root can be represented as $a+b r+$ $c r^{2}$, where $r \in \mathbb{R}$ and $a, b, c, r^{3} \in \mathbb{Q}$. ### 3.3 Representability using only one root of prime order 3.11. Can the following number be represented in the form $$ a_{0}+a_{1} \sqrt[7]{2}+a_{2} \sqrt[7]{2^{2}}+\cdots+a_{6} \sqrt[7]{2^{6}} $$ with $a_{0}, a_{1}, a_{2}, \ldots, a_{6} \in \mathbb{Q}$ : (a) $\sqrt{3}$; (b) $\cos \frac{2 \pi}{21}$; (c) $\sqrt[11]{3}$; (d) $\sqrt[7]{3}$; (e) some root of the polynomial $x^{7}-4 x+2$ ? Answers: no. The arguments are similar to those in the solutions of problems 3.7. Use lemmas stated below. Lemma 3.12. Let $q$ be a prime number, $r \in \mathbb{R}-\mathbb{Q}$ and $r^{q} \in \mathbb{Q}$. (a) Irreducibility. The polynomial $x^{q}-r^{q}$ is irreducible over $\mathbb{Q}$. (b) Linear independence. If $r$ is a root of a polynomial $A$ which degree is less than $q$, then $A=0$. (c) Conjugation. If $r$ is a root of a polynomial, then all the numbers $r \varepsilon_{q}^{k}, k=1,2,3, \ldots, q-1$, are also roots of this polynomial. (d) Rationality. If $A$ is a polynomial, then the number $A(r)$ is a root of some nonzero polynomial which degree is at most $q$. Denote $\mathbb{Q}\left[\varepsilon_{q}\right]:=\left\{a_{0}+a_{1} \varepsilon_{q}+a_{2} \varepsilon_{q}^{2}+\ldots+a_{q-2} \varepsilon_{q}^{q-2}: a_{0}, \ldots, a_{q-2} \in \mathbb{Q}\right\}$. 3.13. Let $q$ be a prime number, $r \in \mathbb{C}-\mathbb{Q}\left[\varepsilon_{q}\right]$ and $r^{q} \in \mathbb{Q}\left[\varepsilon_{q}\right]$. (a) The polynomial $x^{q}-r^{q}$ is irreducible over $\mathbb{Q}\left[\varepsilon_{q}\right]$. (b), (c) Prove the analogues of parts (b,c) of the previous problem for a polynomial with coefficients in $\mathbb{Q}\left[\varepsilon_{q}\right]$. Lemma 3.14.* Let $q$ be a prime number, $r \in \mathbb{R}-\mathbb{Q}$ and $r^{q} \in \mathbb{Q}$. (a) Irreducibility over $\mathbb{Q}\left[\varepsilon_{q}\right]$. The polynomial $x^{q}-r^{q}$ is irreducible over $\mathbb{Q}\left[\varepsilon_{q}\right]$. (b) Linear independence over $\mathbb{Q}\left[\varepsilon_{q}\right]$. If $A$ is a polynomial of degree less than $q$ with coefficients in $\mathbb{Q}\left[\varepsilon_{q}\right]$ and $A(r)=0$, then $A=0$. Theorem 3.15. Assume that a polynomial is irreducible over $\mathbb{Q}$ and has an irrational root $A(r)$, where $A$ is a polynomial and $r \in \mathbb{R}$ is such that $r^{q} \in \mathbb{Q}$ for some prime $q$. Then the polynomial has degree $q$ and, if $q \neq 2$, has no other real roots. The proof is analogous to the proofs of Theorems 2.3, 3.9 and to the solutions of 3.11 (abc). Apply the Conjugation Lemma 3.12.c, the Rationality Lemma 3.12.d, and the Linear Independence over $\mathbb{Q}\left[\varepsilon_{q}\right]$ Lemma 3.14.b. Lemma 3.16 (Extension). The number expressible by real radicals with only one root extraction is equal to $A(r)$ for some $r \in \mathbb{R}, q \in \mathbb{Z}$ and $A \in \mathbb{Q}[x]$, with $r^{q} \in \mathbb{Q}$. The proof is similar to the proof of the Extension Lemma 3.10. 3.17. (a-d) Prove the assertions analogous to Problem 3.12 with $\mathbb{Q}$ replaced by any set $F \subset \mathbb{R}$ which is closed under operations of addition, subtraction, multiplication and division by a non-zero number (and with polynomials over $\mathbb{Q}$ replaced by polynomials over $F$ ). ## Solutions for problems before the semifinal 2.1. Answers: (a), (b), (c), (d) "- yes, (e), (f), (g), (h), (i) "- no. (a), (c) We have $\sqrt{3+2 \sqrt{2}}=\sqrt[3]{7+5 \sqrt{2}}=1+\sqrt{2}$. (b) We have $\frac{1}{7+5 \sqrt{2}}=\frac{7-5 \sqrt{2}}{7^{2}-2 \cdot 5^{2}}=-7+5 \sqrt{2}$. (d) We have $\cos (2 \pi / 5)=(\sqrt{5}-1) / 4$. (e) Assume that $\sqrt[3]{2}$ is representable in this form. Then $$ 2=(\sqrt[3]{2})^{3}=\left(a^{3}+3 a b\right)+\left(3 a^{2}+b\right) \sqrt{b} $$ Since $3 a^{2}+b \neq 0$, we have $\sqrt{b} \in \mathbb{Q}$. Thus $\sqrt[3]{2} \in \mathbb{Q}$, which is a contradiction. Other proofs are similar to the proof of Theorem 2.3. (f) A sketch for the first solution. It is easier to prove right away that $$ \sqrt[3]{2} \neq a+p \sqrt{b}+q \sqrt{c}+r \sqrt{b c}, \quad \text { for any } \quad a, b, c, p, q, r \in \mathbb{Q} $$ It sufficies to show that $\sqrt[3]{2} \neq u+v \sqrt{c}$ for any $u, v, c \in \mathbb{Q}[\sqrt{b}]:=$ $\{x+y \sqrt{b}: x, y \in \mathbb{Q}\}$. The idea of our proof is that numbers from $\mathbb{Q}[\sqrt{b}]$ (with $b$ fixed) are ijas good asi, rational numbers, that is the sum, the difference, the product and the quotient of the numbers from $\mathbb{Q}[\sqrt{b}]$ are also the numbers from $\mathbb{Q}[\sqrt{b}]$ (or, scientifically speaking, $\mathbb{Q}[\sqrt{b}]$ - is a number field). Therefore we can prove the assertion simillary to (e). $A$ sketch for the second solution. Assume that $\sqrt{2}+\sqrt[3]{2}=a+\sqrt{b}$ for some $a, b \in \mathbb{Q}$. This number is a root of the polynomial $P(x)=$ $\left((x-\sqrt{2})^{3}-2\right)\left((x+\sqrt{2})^{3}-2\right)$ having rational coefficients. From (e) it follows that $\sqrt{2}+\sqrt[3]{2} \notin \mathbb{Q}$. Hence, $\sqrt{b} \notin \mathbb{Q}$. By the Conjugation Lemma 2.2 (e) for $r=\sqrt{b}$ we have $P(a-\sqrt{b})=0$. Since $\sqrt{b} \notin \mathbb{Q}$, then roots $a \pm \sqrt{b}$ are different. The polynomial $P$ has only two real roots, namely $\sqrt{2}+\sqrt[3]{2}$ and $-\sqrt{2}+\sqrt[3]{2}$. Thus $a+\sqrt{b}=\sqrt{2}+\sqrt[3]{2}$ and $a-\sqrt{b}=-\sqrt{2}+\sqrt[3]{2}$. Therefore $\sqrt[3]{2}=a \in \mathbb{Q}$. This is a contradiction. (g) Assume that $\cos (2 \pi / 9)$ is representable in this form. By the formula for the cosine of a triple angle $\cos (2 \pi / 9)$ is a root of the equation $4 x^{3}-3 x=-\frac{1}{2}$. By Lemma 2.2 (f) this equation has a rational root, which is a contradiction. Another proof is analogous to Theorem 2.3. (h) The roots of the polynomial $P(x)=\left(x^{2}-2\right)^{2}-2$ are four numbers of the form $\pm \sqrt{2 \pm \sqrt{2}}$, where the signs need not agree. All these numbers are irrational. From Theorem 2.3 it follows that it is sufficient to prove that the polynomial $P$ cannot be written as a product of two quadratic trinomials with rational coefficients. This irreducibility follows from the fact that the product of any two roots of $P$ is irrational. (i) (Here we use the text by I. Braude-Zolotarev.) The equality $$ \cos (2 \pi / 7)+\cos (4 \pi / 7)+\cos (6 \pi / 7)+\ldots+\cos (14 \pi / 7)=0 $$ implies that $\cos (2 \pi / 7)+\cos (4 \pi / 7)+\cos (6 \pi / 7)=-1 / 2$. Applying the formulas $\cos 2 \alpha=2 \cos ^{2} \alpha-1$ and $\cos 3 \alpha=4 \cos ^{3} \alpha-3 \cos \alpha$, we find that $\cos (2 \pi / 7)$ is a root of the equation $8 t^{3}+4 t^{2}-4 t-1=0$. Substituting $u=2 t$ we get $u^{3}+u^{2}-2 u-1=0$. This equation has no rational roots. Hence the same holds for $8 t^{3}+4 t^{2}-4 t-1=0$. Thus the polynomial $8 t^{3}+4 t^{2}-4 t-1=0$ is irreducible over $\mathbb{Q}$. Now the negative answer to the question follows from Lemma 2.2.f. (j) is similar to (f). 2.3. Suppose on the contrary that the given polynomial $P(x)$ has a root $x_{0}=a \pm \sqrt{b}$. By the Conjugation Lemma 2.2.e and analogously to it, the number $x_{1}=a \mp \sqrt{b}$ is also a root of $P$. If $b=0$, then the statement is obvious. So assume that $b \neq 0$. Then $x_{0} \neq x_{1}$. Therefore, $P$ is divisible by $(x-a)^{2}-b$. Since $\operatorname{deg} P>2$ then $P$ is reducible. This is a contradiction. 2.4. Let $\sqrt{c}$ be a number we get with only one extraction of the root, where $c \in \mathbb{Q}$. Prove that all the obtained numbers have the form $a+b \sqrt{c}$ with $a, b \in \mathbb{Q}$. 2.5. Answer: The number is representable if and only if $n \in\{1,2,3,4,5,6,8,10,12\}$. Or, equivalently, $\varphi(n) \in\{1,2,4\}$. 2.8. (a) The polynomial $(x-y)^{2}(y-z)^{2}(z-x)^{2}$ is symmetric. (One may also reduce (a) to (b).) (b) Set $$ M=x^{9} y+y^{9} z+z^{9} x \quad \text { and } \quad N=y^{9} x+x^{9} z+z^{9} y $$ Then $M+N$ and $M N$ are symmetric polynomials. Therefore they are polynomials in elementary symmetric polynomials $\sigma_{1}, \sigma_{2}, \sigma_{3}$. Finally, $M$ itself now can be expressed via $M+N$ and $M N$ by the 'formula for the roots of a quadratic equation', see beginning of $\S 2.3$. 2.9. (a) One possible answer is $u+v \varepsilon_{3}+w \varepsilon_{3}^{2}$. (b) Set $$ \begin{gathered} M=x_{1} x_{3}+x_{3} x_{5}+x_{5} x_{7}+x_{7} x_{9}+x_{9} x_{1} \quad \text { and } \\ N=x_{2} x_{4}+x_{4} x_{6}+x_{6} x_{8}+x_{8} x_{10}+x_{10} x_{2} \end{gathered} $$ Now one can argue as in 2.8.b. 2.10. (a) The square $(x-y)^{2}(x-z)^{2}(x-t)^{2}(y-z)^{2}(y-t)^{2}(z-t)^{2}$ is symmetric, cf. 2.8.a. (b) Set $$ M=x y+z t, \quad N=x z+y t, \quad K=x t+y z $$ By 2.8.c, $M$ can be expressed by radicals using the polynomials $$ M+N+K, \quad M N+M K+N K, \quad M N K $$ Analogously to the solutions of Problems $2.8 \mathrm{c}$ above and 2.10 (d) below, these polynomials are symmetric. Thus $M=x y+z t$ is expressible by radicals. (c) Set $M=(x+y-z-t)^{2}, \quad N=(x+z-y-t)^{2}, \quad K=(x+t-y-z)^{2}$. Then repeat the solution of part (b) to obtain $M=(x+y-z-t)^{2}$. Then it is easy to obtain $x+y-z-t$. Alternative solution. We have $(x+y-z-t)^{2}=\left(x^{2}+y^{2}+z^{2}+t^{2}\right)+2(x y+t z)-2(x t+y z)-2(x z+y t)$. The first summand is symmetric and the other summands are expressible by radicals due to part (b). Thus $x+y-z-t$ is expressible by radicals. Solution of the equation of fourth degree using Lagrange resolvent method (solution of the problem 2.10(d)). To find the roots $x, y, z, t$ of the fourth degree equation it is enough to find the expressions for $a, b, c, d$ from Problem 2.11 (a). By Vieta theorem, $a$ is the coefficient of the equation. Under substitution $x \leftrightarrow y$, polynomials $c^{2}$ and $d^{2}$ are interchanged, and $b^{2}$ goes to itself. After cyclic permutation $x \rightarrow$ $y \rightarrow z \rightarrow t \rightarrow x$, polynomials $b^{2}$ and $d^{2}$ are interchanged, and $c^{2}$ goes to itself. Therefore polynomials $b^{2}, c^{2}, d^{2}$ are permuted for every permutation of variables $x, y, z, t$. Hence their Vieta polynomials, i.e. $$ b^{2}+c^{2}+d^{2}, \quad b^{2} c^{2}+b^{2} d^{2}+c^{2} d^{2}, \quad b^{2} c^{2} d^{2} $$ are symmetric. Then these polynomials in $x, y, z$ can be represented as polynomials in the coefficients of the equation. Now by solving the cubic equation we can get $b^{2}, c^{2}, d^{2}$. Then it is easy to obtain $b, c, d$. 2.11. Repeatedly use the identities $1+\varepsilon+\varepsilon^{2}=0$ and $1+i+i^{2}+i^{3}=$ 0. 2.12. Here we show the solution for $q=5$. (a) We have $$ t\left(\varepsilon_{5} \vec{u}_{\alpha}\right)=t\left(u_{\alpha(5)}, u_{\alpha(1)}, u_{\alpha(2)}, u_{\alpha(3)}, u_{\alpha(4)}\right)=t\left(\vec{u}_{\alpha \circ(54321)}\right) $$ Hence $$ \begin{aligned} & Q\left(\varepsilon_{5} u_{1}, \ldots, \varepsilon_{5} u_{5}, y\right)=\prod_{\alpha \in \Sigma_{5}}\left(y-t\left(\varepsilon_{5} \vec{u}_{\alpha}\right)\right)= \\ & =\prod_{\alpha \in \Sigma_{5}}\left(y-t\left(\vec{u}_{\alpha \circ(54321)}\right)=Q\left(u_{1}, \ldots, u_{5}, y\right)\right. \end{aligned} $$ Here - $(54321) \in \Sigma_{5}$ is the cycle that sends 5 to 4,4 to $3, \ldots, 1$ to 5 . - the last equality holds because when $\alpha$ ranges through $\Sigma_{5}$, so does $\alpha \circ(54321)$. (b) There is a homogeneous polynomial $P_{k} \in \mathbb{Q}\left[\varepsilon_{5}\right]\left[u_{1}, \ldots, u_{5}\right]$ (of 'degree' $120-k)$ such that the coefficient of $y^{k}$ in $Q$ is $P\left(u_{1}, \ldots, u_{5}\right)$, i.e. $$ Q\left(u_{1}, \ldots, u_{5}, y\right)=\sum_{k=0}^{120} P_{k}\left(u_{1}, \ldots, u_{5}\right) y^{k} $$ By (a) and by homogeneity we have $$ P_{k}\left(u_{1}, \ldots, u_{5}\right)=P_{k}\left(\varepsilon_{5} u_{1}, \ldots \varepsilon_{5} u_{5}\right)=\varepsilon_{5}^{-k} P_{k}\left(u_{1}, \ldots, u_{5}\right) $$ If $k$ is not divisible by 5 , we obtain $P_{k}\left(u_{1}, \ldots, u_{5}\right)=0$ as required. (c) The polynomial $Q\left(u_{1}, \ldots, u_{5}, y\right)$ is symmetric in $u_{1}, \ldots, u_{5}$. So all the coefficients $\left(P_{k}\right.$ from (b)) of the corresponding polynomial from $\mathbb{Q}\left[\varepsilon_{5}, u_{1}, \ldots, u_{5}\right][y]$ are symmetric in $u_{1}, \ldots, u_{5}$. Now the statement follows from the fundamental theorem on symmetric polynomials, Vieta theorem and the fact that the coefficients of $f$ are rational. 2.14. (a) We will prove that there exists $\alpha \in \mathbb{R}$ such that $f(x, y)=$ $\alpha(x-y)$. Since the polynomial $f^{2}=p$ is symmetric, we can assume that the polynomial $q$ is linear in third variable, i.e. $q(u, v, w)=a(u, v)+$ $b(u, v) w$ for some $a, b \in \mathbb{R}[u, v]$ (otherwise we can change $q$ while preserving $f, p)$. Then we have $x=a(x+y, x y)+b(x+y, x y) f(x, y)$. Now we get $p b^{2}=f^{2} b^{2}=(x-a)^{2}=(y-a)^{2}$. By Lemma 2.15.b we get $x-a=a-y$, since the case $x-a=y-a$ is impossible. Hence $a=(x+y) / 2$. Then $(x-y)^{2}=4 f^{2} b^{2}=4 p b^{2}$. If the polynomial $p=f^{2}$ is constant, the polynomial $b= \pm(x-y) / 2 \sqrt{p}$ is not symmetric. Therefore $p$ is not constant. Thus $b$ is constant. Hence $2 x=2 q=$ $x+y+2 b f$, from which $b \neq 0$ and $f=\alpha(x-y)$ for $\alpha=1 / 2 b$. (b) We will prove that $k$ is even and that there exists $\alpha \in \mathbb{R}$ such that $f(x, y)=\alpha(x-y)$. We can use induction on $k$ with the application of part (a) and the generalization of Lemmas 2.15.be, 2.16. If $k$ is odd, from Lemma 2.16.b we get that $f$ is symmetric. That contradicts the equality $x=q(x+y, x y, f(x, y))$. If $k=4$, then $f^{2}$ is either symmetric or antisymmetric. The first case reduces to part (a). The second one gives us $f^{2}(x, y)+f^{2}(y, x)=0$. We solve the case of arbitrary even $k$ analogously. (c) Analogously to part (a) we get $x=a+b f$. Therefore, $f$ is a rational fraction. Now the solution is analogous to part (a). 2.15. (a) Define the leading term of a polynomial so that the leading term of the product is equal to the product of the leading terms of the factors. (b) This follows from part (a). (c) We have $f^{2}+f g+g^{2}=\left(f+\frac{g}{2}\right)^{2}+\frac{3}{4} g^{2}=\left(f-\varepsilon_{3} g\right)\left(f-\varepsilon_{3}^{2} g\right)$. (d) This follows from part (c). (e) This follows from part (f). (f) Prove and apply the Bezout theorem for polynomials in $u$ with coefficients in $\mathbb{R}[v]$. 2.16. (a) Since $f^{2}$ is symmetric, we have $f(x, y)^{2}=f(y, x)^{2}$. Now by the statement 2.15.b we have $f(x, y)= \pm f(y, x)$. (b) Use the analogues of statements 2.15.ce. (c) See the hint for 2.15.f. 2.17. Answer: 2.15.abf, 2.16.abc. 2.22. For $n=3$, the set of polynomials expressible by real radicals is contained in the set of cyclic symmetric polynomials. This statement can be proved by induction on the number of operations from the definition of expressibility in radicals. The induction step follows from Lemma 2.23 on the preservation of cyclic symmetry. Since the polynomial $x$ is not cyclic symmetric, it is also not expressible in real radicals. 2.23. The proof can be found in [Sk19, p. 9.4.2]. 2.24. Answer: 2.19 .abcd, 2.20 .b, 2.23 for all $q$ which are not divisible by 3 . ## Toward algorithms of solving algebraic equations presented by A. Enne, A. Skopenkov, A. Glebov, A. Chilikov, B. Vukorepa ## Solutions for problems after the semifinal 3.1. (b) Use the analog of Problem 3.2.c for $n=3$. 3.2. Answer: (c) - true, (a), (b), (d) - not true. (a) See 2.9.a. (b) Consider the polynomial $x_{1}+\varepsilon_{5} x_{2}+\varepsilon_{5}^{2} x_{3}+\varepsilon_{5}^{3} x_{4}+\varepsilon_{5}^{4} x_{5}$. (d) Consider the polynomial $\prod_{i0, P(1)<0$, and $P(2)>0$, the polynomial $P$ has a real root $x_{k}$ different from $x_{0}$. We have $\overline{\varepsilon_{7}^{k}}=\varepsilon_{7}^{-k}$. Hence $x_{k}=\overline{x_{k}}=x_{7-k}$. A contradiction. Second solution. Denote by $P$ the polynomial such that $\cos 7 x=$ $P(\cos x)$ (prove that it exists!). The roots of the polynomial $2 P(x)+1$ are real numbers $y_{k}=\cos \frac{2(3 k+1) \pi}{21}$ with $k=0, \ldots, 6$. One of them, namely $y_{2}=-1 / 2$, is rational. In the following paragraph we prove that $y_{0}$ is irrational. (Otherwise, the equality $\varepsilon_{21}^{2}-2 y_{0} \varepsilon_{21}+1=0$ implies that $\varepsilon_{21}=$ $a+i \sqrt{b}$ for some $a, b \in \mathbb{Q}$. Then the number $\varepsilon_{7}=\varepsilon_{21}^{3}$ also has this form. But $\varepsilon_{7}$ is a root of the irreducible ${ }^{4}$ polynomial $1+x+\cdots+x^{6}$,[^13]which contradicts to the analogue of Theorem 2.3 for numbers of the form $a+i \sqrt{b}$.) Thus the number $y_{0}$ is an irrational root of the polynomial $\frac{2 P(x)+1}{2 x+1}$ which has degree 6. Then Conjugation Lemma 3.12.c and Linear Independence over $\mathbb{Q}\left[\varepsilon_{q}\right]$ Lemma 3.14.b show that this polynomial has seven distinct roots, which is impossible. (c) Assume that $\sqrt[11]{3}$ is representable in this form. Then by the Rationality Lemma 3.12 (d), there exists a nonzero polynomial of degree at most 7 having $\sqrt[11]{3}$ as a root. This contradicts the irreducibility of the polynomial $x^{11}-3$ over $\mathbb{Q}$. (d) Assume that $\sqrt[7]{3}$ is representable in this form. Analogously to (a), all the complex roots of the polynomial $x^{7}-3$ are $A\left(r \varepsilon_{7}^{k}\right)$ for $k=$ $0,1,2, \ldots, 6$. Therefore, $A(r) \varepsilon_{7}^{s}=A\left(r \varepsilon_{7}\right)$ for some $s \in\{1,2,3,4,5,6\}$. Hence by the Linear Independence Lemma over $\mathbb{Q}\left[\varepsilon_{q}\right] 3.14$ (b) we have $a_{k}=0$ for each $k \neq s$. Therefore, $\sqrt[7]{3}=a_{s} r^{s}$. This is a contradiction. (e) Assume that one of the roots is representable in this form. The given polynomial $P$ has no rational roots. Then Conjugation Lemma 3.12.c and Linear Independence over $\mathbb{Q}\left[\varepsilon_{q}\right]$ Lemma 3.14.b yield that $P$ has pairwise distinct roots $x_{k}:=A\left(r \varepsilon_{7}^{k}\right)$ for $k=0,1,2, \ldots, 6$. Since $P(0)>0, P(1)<0$, and $P(2)>0$, the polynomial $P$ has a real root $x_{k}$ distinct from $x_{0}$. From the equality $\overline{\varepsilon_{7}^{k}}=\varepsilon_{7}^{-k}$ it follows that $x_{k}=\overline{x_{k}}=x_{7-k}$. This is a contradiction. 3.12. (a) All the roots of the polynomial $x^{q}-r^{q}$ are $r, r \varepsilon_{q}, r \varepsilon_{q}^{2}, \ldots, r \varepsilon_{q}^{q-1}$. Assume that $x^{q}-r^{q}$ is reducible over $\mathbb{Q}$. Then the absolute value of the constant term of one of its unitary irreducible factors is rational and equals to the product of absolute values of $k$ of these roots, $0\operatorname{deg} P$ we get a contradiction. When $q=\operatorname{deg} P$ the conditions $q \neq 2$ and $\overline{x_{k}}=x_{q-k} \neq x_{k}$ yield the uniqueness of the real root. ## References [Al] Alekseev V. B., Abel's Theorem in Problems and Solutions. Springer Netherlands, 2004. [AB] Akhtyamov D., Bogdanov I., Solvability of cubic and quartic equations using one radical. http://arxiv.org/abs/1411. 4990. [Dor] Dörrie H., 100 Great Problems of Elementary Mathematics: Their History and Solution. New York: Dover Publ, 1965. [E2] Edwards H. M., The construction of solvable polynomials Bull. Amer. Math. Soc. 2009. V. 46. P 397-411. Errata: Bull. Amer. Math. Soc. 46 (2009), 703-704. [Es] Esterov A., Galois theory for general systems of polynomial equations, https://arxiv.org/abs/1801.08260 [FT] Fuchs D., Tabachnikov S., Mathematical Omnibus. AMS, 2007. [Had] Hadlock Ch. R., Field Theory and its Classical Problems. The Mathematical Association of America, 1978. (Carus Mathematical Monographs, N 19.) [Ka] Kannunikov A. L., The beginning of Galois theory: solvability of algebraic equations by radicals (in Russian). http://www.mathnet.ru/conf 1015. [Ko17] Kogan E., Set complexity of construction of a regular polygon, https://arxiv.org/abs/1711. 05807. [Kol] Kolosov V. A., Theorems and problems in algebra, number theory and combinatorics (in Russian). M.: Helios, 2001. [Ler] Lerner L., Galois Theory without abstract algebra. http://arxiv.org/abs/1108.4593. [Pr07-2] Prasolov V. V., Problems in algebra, arithmetics and analysis, Moscow, MCCME, 2007. [PSo] Prasolov V. V., Solovyev Y. P., Elliptic Functions and Elliptic Integrals. AMS, 1997. [Saf] Safin A., A program for construction of regular polygons by compass and ruler. http://www.mccme.ru/mmks/dec08/ Safin.pdf. [Sk10] Skopenkov A., Basic embeddings and Hilbert's 13th problem (in Russian), Mat. Prosveschenie, 14 (2010) 143-174, http: //arxiv.org/abs/1001.4011. Abridged English translation: http://arxiv.org/abs/1003.1586. [Sk11] Skopenkov A., A simple proof of the Abel-Ruffini theorem on insolvability of equations in radicals (in Russian), Mat. Prosveschenie, 15 (2011) 113-126. http://arxiv.org/abs/ 1102.2100. [Sk15] Skopenkov A., A short elementary proof of the Ruffini-Abel Theorem. http://arxiv.org/abs/1508.03317. [Sk19] Skopenkov A., Mathematics via problems: from olympiads and math circles to a profession. Algebra. AMS, Providence, to appear. [St94] Stillwell J., Galois theory for beginners, Amer. Math. Monthly, 101 (1994), 22-27. [T] Tikhomirov V. M., Abel and his great theorem, Kvant. 2003. N1. P. 11-15. [Vag] Vaguten H., Conjugated numbers (in Russian), Kvant. 1980. N 2. P. 26-32. [Vi] Vinberg E. B., Algebra of polynomials (in Russian). M.: Prosveschenie, 1980. [ZSS] Mathematics via problems: from olympiads and math circles to a profession (in Russian). Editors: A. Zaslavsky, A. Skopenkov and M. Skopenkov. MCCME, 2018. Abridged version: http://www.mccme.ru/circles/oim/materials/ sturm.pdf. [SZ19] Mathematics via problems: from olympiads and math circles to a profession. Geometry and Combinatorics. Editors: A. Zaslavsky, and M. Skopenkov. AMS, Providence, to appear. [W] Van der Waerden B. L., Algebra. Frederick Ungar Publishing, 1970. ## 4 Additional problems for successful teams 4.1. (a) Let $x, y, r \in \mathbb{R}, p, g \in \mathbb{Q}[u, v]$ and $p_{1} \in \mathbb{Q}[u, v, w]$ be such that $g(x, y) \notin \mathbb{Q}(x+y, x y)$ and $$ \left\{\begin{array}{l} r^{2}=p(x+y, x y) \\ g(x, y)=p_{1}(x+y, x y, r) \end{array}\right. $$ (cf. Problem 2.14.c). Then $r \in \mathbb{Q}(x, y)$. (b) Let $x, y, r \in \mathbb{R}, p \in \mathbb{Q}[\sqrt{2}][u, v], g \in \mathbb{Q}[u, v]$ and $p_{1} \in \mathbb{Q}[\sqrt{2}][u, v, w]$ be such that $g(x, y) \notin \mathbb{Q}(x+y, x y, \sqrt{2})$ and the equations of (a) hold. Then there are $\rho \in \mathbb{Q}(x, y), \pi \in \mathbb{Q}[\sqrt{2}][u, v]$ and $\pi_{1} \in \mathbb{Q}[\sqrt{2}][u, v, w]$ such that the equations of (a) hold with $r, p, p_{1}$ replaced by $\rho, \pi, \pi_{1}$. (c) Rationalization Lemma. Let $x, y, r \in \mathbb{R}$ and $F \subset \mathbb{R}$ a field containing $x+y, x y, r^{2}$ but not $r$. If $F(r) \cap \mathbb{Q}(x, y) \not \subset F$, then there is $\rho \in \mathbb{Q}(x, y)$ such that $\rho^{2} \in F$ and $F(\rho)=F(r)$. 4.2. Denote $a_{j}=\sigma_{j}\left(x_{1}, x_{2}, x_{3}\right), j=1,2,3$. (a) Let $x_{1}, x_{2}, x_{3}, r \in \mathbb{R}, p, g \in \mathbb{Q}\left[u_{1}, u_{2}, u_{3}\right]$ and $p_{1} \in \mathbb{Q}\left[u_{1}, u_{2}, u_{3}, v\right]$ be such that $g\left(x_{1}, x_{2}, x_{3}\right) \notin \mathbb{Q}\left(a_{1}, a_{2}, a_{3}\right)$ and $$ \left\{\begin{array}{l} r^{2}=p\left(a_{1}, a_{2}, a_{3}\right) \\ g\left(x_{1}, x_{2}, x_{3}\right)=p_{1}\left(a_{1}, a_{2}, a_{3}, r\right) \end{array}\right. $$ Then $r \in \mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right)$. (b) Rationalization Lemma. Let $x_{1}, x_{2}, x_{3}, r \in \mathbb{R}$ and $F \subset \mathbb{R}$ a field containing $a_{1}, a_{2}, a_{3}, r^{2}$ but not $r$. If $F(r) \cap \mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right) \not \subset F$, then there is $\rho \in \mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right)$ such that $\rho^{2} \in F$ and $F(\rho)=F(r)$. (c) Proposition. If $x_{1}, x_{2}, x_{3} \in \mathbb{R}$ and $x_{1}$ is $\left\{a_{1}, a_{2}, a_{3}\right\}$-expressible by quadratic real radicals, then $x_{1}$ is $\left\{a_{1}, a_{2}, a_{3}\right\}$-expressible by quadratic real radicals so that every radical is in $\mathbb{Q}\left(x_{1}, x_{2}, x_{3}\right)$. 4.3. (a) Let $x, y, r \in \mathbb{C}, p \in \mathbb{Q}[u, v]$ and $p_{1} \in \mathbb{Q}[u, v, w]$ be such that $$ \left\{\begin{array}{l} r^{3}=p(x+y, x y) \\ x=p_{1}(x+y, x y, r) \end{array}\right. $$ (cf. Problem 2.14.d for $k=3$ ). Then $r \in \mathbb{Q}\left[\varepsilon_{3}\right](x, y)$. (b) Same as (a) with $x=p_{1}(x+y, x y, r)$ replaced by $g(x, y)=$ $p_{1}(x+y, x y, r)$ for some $g \in \mathbb{Q}[u, v]$ such that $g(x, y) \notin \mathbb{Q}(x+y, x y)$. (c) Rationalization Lemma. Let $x, y, r \in \mathbb{C}$ and $F \subset \mathbb{C}$ a field containing $x+y, x y, \varepsilon_{3}, r^{3}$ but not $r$. If $F(r) \cap \mathbb{Q}(x, y) \not \subset F$, then there is $\rho \in \mathbb{Q}(x, y)$ such that $\rho^{3} \in F$ and $F(\rho)=F(r)$. (d) Rationalization Lemma. Same as (c) with $x, y$ replaced by $x_{1}, \ldots, x_{n}$ and $x+y, x y$ replaced by $\sigma_{1}\left(x_{1}, \ldots, x_{n}\right), \ldots, \sigma_{n}\left(x_{1}, \ldots, x_{n}\right)$. (e) Rationalization Lemma. Same as (d) with $r^{3}, \rho^{3}$ replaced by $r^{q}, \rho^{q}$ for a prime $q$ and $\varepsilon_{3}$ replaced by $\varepsilon_{q}$. (f) Proposition. If $$ x_{1}, \ldots, x_{n} \in \mathbb{C}, \quad M:=\left\{\sigma_{1}\left(x_{1}, \ldots, x_{n}\right), \ldots, \sigma_{n}\left(x_{1}, \ldots, x_{n}\right)\right\} $$ and $x_{1}$ is $M$-expressible by radicals, then $x_{1}$ is $M$-expressible by radicals so that every radical is in $\bigcup_{q=3}^{\infty} \mathbb{Q}\left[\varepsilon_{q}\right]\left(x_{1}, \ldots, x_{n}\right)$. 4.4. There are numbers $x, y \in \mathbb{R}$ such that if $p \in \mathbb{Q}[u, v]$ and $p(x, y)=0$, then $p=0$. Such numbers are called algebraically independent over $\mathbb{Q}$. [^0]: *Мы также хотим поблагодарить Михаила Скопенкова, Илью Богданова, Кейта Конрада за существенную помощь в подготовке проекта. [^1]: ${ }^{*}$ We also wish to thank Mikhail Skopenkov, Ilya Bogdanov and Keith Conrad for a variety of useful comments on this project. [^2]: ${ }^{1}$ Авторы благодарят Д.Крекова за помощь в подготовке проекта [^3]: ${ }^{1}$ The authors are grateful to D.Krekov for the help in the preparing of the project [^4]: «Треугольным деревом» назовем граф, определеямый по индукции следующим образом. Самое маленькое треугольное дерево вопреки своему названию - это полный граф на двух вершинах. Если уже задано какое-нибудь треугольное дерево, мы можем взять любое ребро $A B$ в нем, взять новую вершину $C$ и добавить к дереву вершину $C$ и ребра $A C, B C$. Помеченным треугольным деревом будем называть треугольное дерево, у которого вершины пронумерованы числами от 1 до $n$. [^5]: $1_{\text {на самом }}$ деле, для непериодических замощений - никогда, и позже мы это докажем [^6]: ${ }^{1}$ Further we will prove that for aperiodic tilings it is never possible. [^7]: "Благодарим Я. Абрамова, Д. Елисеева, А. Канунникова, О. Орел, А. Петухова, Н. Хорошавкину, Г. Челнокова и жюри ЛКТГ за обсуждения и перевод частей текста. Б. Вукорепа: Загребский Университет. А. Глебов: Новосибирский Государственный Университет. A. Еннэ: Петрозаводский Государственный Университет. $A$. Скопенков: Московский Физико-Технический Институт, Независимый Московский Университет. http://www.mccme.ru/ skopenko. А. Чиликов: Московский Государственный Технический Университет им. Баумана, Московский ФизикоТехнический Институт. ${ }^{1}$ Из §1 далее используется только §1.5, можно сразу перейти к нему и решению задач. [^8]: ${ }^{3}$ Это деление с остатком - то же самое, что «замена» $x^{2}$ на $r^{2}$. [^9]: ${ }^{4}$ Неприводимость многочлена $g(x)=1+x+\ldots+x^{6}$ можно показать, например, применив признак Эйзенштейна [ZSS, п. 5.5.2] к многочлену $g(x+1)$. Впрочем, здесь достаточно доказать, что у него нет рациональных делителей степени 1 и 2. [^10]: ${ }^{*}$ We are grateful to Ya. Abramov, G. Chelnokov, D. Eliseev, A. Kanunnikov, N. Khoroshavkina, O. Orel, A. Petukhov and the jury of SCTT for useful discussions and for translation of some parts of the text. A. Enne: Petrozavodsk State University. A. Chilikov: Bauman Moscow State Technical University, Moscow Institute of Physics and Technology. A. Glebov: Novosibirsk State University. A. Skopenkov: Moscow Institute of Physics and Technology, Independent University of Moscow. http://www.mccme.ru/ skopenko. B. Vukorepa: University of Zagreb. ${ }^{1}$ From $\S 1$ in what follows we use only $\S 1.5$, so you can read that subsection and start solving problems. [^11]: ${ }^{2}$ See e.g. [Dor, § 25], [Pr07-2, appendix 8], [FT, Lecture 5], [ZSS, §5], [Dor, St94, Kol, Ler, T, Sk11, Sk15] and this text. Exposition in [Al] is closer to this 'direct' style. Large part of $[\mathrm{Al}]$ contains theory not required to prove the weak version of Abel theorem announced as the main result, see [Sk15, end of remark 7]. However, the author of $[\mathrm{Al}]$ succeeded in avoiding unmotivated exposition of the most complicated part of the theory. The proof from $[\mathrm{Al}]$ is exposed a shorter and easier way in [FT, Lecture 5], [Sk11]. Note that proofs in many of these sources are incomplete. See [ZSS, footnote 12 in p. 113 and end of §5.5.4], [Sk15, Discussion]. In spite of these drawbacks the above elementary expositions were more useful to us than formal expositions (in standard textbooks intended for the theory) which start with several hundreds pages of definitions and results whose role in the proof of the insolvability theorem is not clear at the moment of their statements. [^12]: ${ }^{3}$ The division with a reminder is equivalent to 'replacing' $x^{2}$ by $r^{2}$. [^13]: ${ }^{4}$ The irreducibility of the polynomial $g(x)=1+x+\ldots+x^{6}$ can be proved, e.g.,