Stay ahead of learning milestones! Enroll in a class over the summer!

G
Topic
First Poster
Last Poster
k a May Highlights and 2025 AoPS Online Class Information
jlacosta   0
May 1, 2025
May is an exciting month! National MATHCOUNTS is the second week of May in Washington D.C. and our Founder, Richard Rusczyk will be presenting a seminar, Preparing Strong Math Students for College and Careers, on May 11th.

Are you interested in working towards MATHCOUNTS and don’t know where to start? We have you covered! If you have taken Prealgebra, then you are ready for MATHCOUNTS/AMC 8 Basics. Already aiming for State or National MATHCOUNTS and harder AMC 8 problems? Then our MATHCOUNTS/AMC 8 Advanced course is for you.

Summer camps are starting next month at the Virtual Campus in math and language arts that are 2 - to 4 - weeks in duration. Spaces are still available - don’t miss your chance to have an enriching summer experience. There are middle and high school competition math camps as well as Math Beasts camps that review key topics coupled with fun explorations covering areas such as graph theory (Math Beasts Camp 6), cryptography (Math Beasts Camp 7-8), and topology (Math Beasts Camp 8-9)!

Be sure to mark your calendars for the following upcoming events:
[list][*]May 9th, 4:30pm PT/7:30pm ET, Casework 2: Overwhelming Evidence — A Text Adventure, a game where participants will work together to navigate the map, solve puzzles, and win! All are welcome.
[*]May 19th, 4:30pm PT/7:30pm ET, What's Next After Beast Academy?, designed for students finishing Beast Academy and ready for Prealgebra 1.
[*]May 20th, 4:00pm PT/7:00pm ET, Mathcamp 2025 Qualifying Quiz Part 1 Math Jam, Problems 1 to 4, join the Canada/USA Mathcamp staff for this exciting Math Jam, where they discuss solutions to Problems 1 to 4 of the 2025 Mathcamp Qualifying Quiz!
[*]May 21st, 4:00pm PT/7:00pm ET, Mathcamp 2025 Qualifying Quiz Part 2 Math Jam, Problems 5 and 6, Canada/USA Mathcamp staff will discuss solutions to Problems 5 and 6 of the 2025 Mathcamp Qualifying Quiz![/list]
Our full course list for upcoming classes is below:
All classes run 7:30pm-8:45pm ET/4:30pm - 5:45pm PT unless otherwise noted.

Introductory: Grades 5-10

Prealgebra 1 Self-Paced

Prealgebra 1
Tuesday, May 13 - Aug 26
Thursday, May 29 - Sep 11
Sunday, Jun 15 - Oct 12
Monday, Jun 30 - Oct 20
Wednesday, Jul 16 - Oct 29

Prealgebra 2 Self-Paced

Prealgebra 2
Wednesday, May 7 - Aug 20
Monday, Jun 2 - Sep 22
Sunday, Jun 29 - Oct 26
Friday, Jul 25 - Nov 21

Introduction to Algebra A Self-Paced

Introduction to Algebra A
Sunday, May 11 - Sep 14 (1:00 - 2:30 pm ET/10:00 - 11:30 am PT)
Wednesday, May 14 - Aug 27
Friday, May 30 - Sep 26
Monday, Jun 2 - Sep 22
Sunday, Jun 15 - Oct 12
Thursday, Jun 26 - Oct 9
Tuesday, Jul 15 - Oct 28

Introduction to Counting & Probability Self-Paced

Introduction to Counting & Probability
Thursday, May 15 - Jul 31
Sunday, Jun 1 - Aug 24
Thursday, Jun 12 - Aug 28
Wednesday, Jul 9 - Sep 24
Sunday, Jul 27 - Oct 19

Introduction to Number Theory
Friday, May 9 - Aug 1
Wednesday, May 21 - Aug 6
Monday, Jun 9 - Aug 25
Sunday, Jun 15 - Sep 14
Tuesday, Jul 15 - Sep 30

Introduction to Algebra B Self-Paced

Introduction to Algebra B
Tuesday, May 6 - Aug 19
Wednesday, Jun 4 - Sep 17
Sunday, Jun 22 - Oct 19
Friday, Jul 18 - Nov 14

Introduction to Geometry
Sunday, May 11 - Nov 9
Tuesday, May 20 - Oct 28
Monday, Jun 16 - Dec 8
Friday, Jun 20 - Jan 9
Sunday, Jun 29 - Jan 11
Monday, Jul 14 - Jan 19

Paradoxes and Infinity
Mon, Tue, Wed, & Thurs, Jul 14 - Jul 16 (meets every day of the week!)

Intermediate: Grades 8-12

Intermediate Algebra
Sunday, Jun 1 - Nov 23
Tuesday, Jun 10 - Nov 18
Wednesday, Jun 25 - Dec 10
Sunday, Jul 13 - Jan 18
Thursday, Jul 24 - Jan 22

Intermediate Counting & Probability
Wednesday, May 21 - Sep 17
Sunday, Jun 22 - Nov 2

Intermediate Number Theory
Sunday, Jun 1 - Aug 24
Wednesday, Jun 18 - Sep 3

Precalculus
Friday, May 16 - Oct 24
Sunday, Jun 1 - Nov 9
Monday, Jun 30 - Dec 8

Advanced: Grades 9-12

Olympiad Geometry
Tuesday, Jun 10 - Aug 26

Calculus
Tuesday, May 27 - Nov 11
Wednesday, Jun 25 - Dec 17

Group Theory
Thursday, Jun 12 - Sep 11

Contest Preparation: Grades 6-12

MATHCOUNTS/AMC 8 Basics
Friday, May 23 - Aug 15
Monday, Jun 2 - Aug 18
Thursday, Jun 12 - Aug 28
Sunday, Jun 22 - Sep 21
Tues & Thurs, Jul 8 - Aug 14 (meets twice a week!)

MATHCOUNTS/AMC 8 Advanced
Sunday, May 11 - Aug 10
Tuesday, May 27 - Aug 12
Wednesday, Jun 11 - Aug 27
Sunday, Jun 22 - Sep 21
Tues & Thurs, Jul 8 - Aug 14 (meets twice a week!)

AMC 10 Problem Series
Friday, May 9 - Aug 1
Sunday, Jun 1 - Aug 24
Thursday, Jun 12 - Aug 28
Tuesday, Jun 17 - Sep 2
Sunday, Jun 22 - Sep 21 (1:00 - 2:30 pm ET/10:00 - 11:30 am PT)
Monday, Jun 23 - Sep 15
Tues & Thurs, Jul 8 - Aug 14 (meets twice a week!)

AMC 10 Final Fives
Sunday, May 11 - Jun 8
Tuesday, May 27 - Jun 17
Monday, Jun 30 - Jul 21

AMC 12 Problem Series
Tuesday, May 27 - Aug 12
Thursday, Jun 12 - Aug 28
Sunday, Jun 22 - Sep 21
Wednesday, Aug 6 - Oct 22

AMC 12 Final Fives
Sunday, May 18 - Jun 15

AIME Problem Series A
Thursday, May 22 - Jul 31

AIME Problem Series B
Sunday, Jun 22 - Sep 21

F=ma Problem Series
Wednesday, Jun 11 - Aug 27

WOOT Programs
Visit the pages linked for full schedule details for each of these programs!


MathWOOT Level 1
MathWOOT Level 2
ChemWOOT
CodeWOOT
PhysicsWOOT

Programming

Introduction to Programming with Python
Thursday, May 22 - Aug 7
Sunday, Jun 15 - Sep 14 (1:00 - 2:30 pm ET/10:00 - 11:30 am PT)
Tuesday, Jun 17 - Sep 2
Monday, Jun 30 - Sep 22

Intermediate Programming with Python
Sunday, Jun 1 - Aug 24
Monday, Jun 30 - Sep 22

USACO Bronze Problem Series
Tuesday, May 13 - Jul 29
Sunday, Jun 22 - Sep 1

Physics

Introduction to Physics
Wednesday, May 21 - Aug 6
Sunday, Jun 15 - Sep 14
Monday, Jun 23 - Sep 15

Physics 1: Mechanics
Thursday, May 22 - Oct 30
Monday, Jun 23 - Dec 15

Relativity
Mon, Tue, Wed & Thurs, Jun 23 - Jun 26 (meets every day of the week!)
0 replies
jlacosta
May 1, 2025
0 replies
k i Adding contests to the Contest Collections
dcouchman   1
N Apr 5, 2023 by v_Enhance
Want to help AoPS remain a valuable Olympiad resource? Help us add contests to AoPS's Contest Collections.

Find instructions and a list of contests to add here: https://artofproblemsolving.com/community/c40244h1064480_contests_to_add
1 reply
dcouchman
Sep 9, 2019
v_Enhance
Apr 5, 2023
k i Zero tolerance
ZetaX   49
N May 4, 2019 by NoDealsHere
Source: Use your common sense! (enough is enough)
Some users don't want to learn, some other simply ignore advises.
But please follow the following guideline:


To make it short: ALWAYS USE YOUR COMMON SENSE IF POSTING!
If you don't have common sense, don't post.


More specifically:

For new threads:


a) Good, meaningful title:
The title has to say what the problem is about in best way possible.
If that title occured already, it's definitely bad. And contest names aren't good either.
That's in fact a requirement for being able to search old problems.

Examples:
Bad titles:
- "Hard"/"Medium"/"Easy" (if you find it so cool how hard/easy it is, tell it in the post and use a title that tells us the problem)
- "Number Theory" (hey guy, guess why this forum's named that way¿ and is it the only such problem on earth¿)
- "Fibonacci" (there are millions of Fibonacci problems out there, all posted and named the same...)
- "Chinese TST 2003" (does this say anything about the problem¿)
Good titles:
- "On divisors of a³+2b³+4c³-6abc"
- "Number of solutions to x²+y²=6z²"
- "Fibonacci numbers are never squares"


b) Use search function:
Before posting a "new" problem spend at least two, better five, minutes to look if this problem was posted before. If it was, don't repost it. If you have anything important to say on topic, post it in one of the older threads.
If the thread is locked cause of this, use search function.

Update (by Amir Hossein). The best way to search for two keywords in AoPS is to input
[code]+"first keyword" +"second keyword"[/code]
so that any post containing both strings "first word" and "second form".


c) Good problem statement:
Some recent really bad post was:
[quote]$lim_{n\to 1}^{+\infty}\frac{1}{n}-lnn$[/quote]
It contains no question and no answer.
If you do this, too, you are on the best way to get your thread deleted. Write everything clearly, define where your variables come from (and define the "natural" numbers if used). Additionally read your post at least twice before submitting. After you sent it, read it again and use the Edit-Button if necessary to correct errors.


For answers to already existing threads:


d) Of any interest and with content:
Don't post things that are more trivial than completely obvious. For example, if the question is to solve $x^{3}+y^{3}=z^{3}$, do not answer with "$x=y=z=0$ is a solution" only. Either you post any kind of proof or at least something unexpected (like "$x=1337, y=481, z=42$ is the smallest solution). Someone that does not see that $x=y=z=0$ is a solution of the above without your post is completely wrong here, this is an IMO-level forum.
Similar, posting "I have solved this problem" but not posting anything else is not welcome; it even looks that you just want to show off what a genius you are.

e) Well written and checked answers:
Like c) for new threads, check your solutions at least twice for mistakes. And after sending, read it again and use the Edit-Button if necessary to correct errors.



To repeat it: ALWAYS USE YOUR COMMON SENSE IF POSTING!


Everything definitely out of range of common sense will be locked or deleted (exept for new users having less than about 42 posts, they are newbies and need/get some time to learn).

The above rules will be applied from next monday (5. march of 2007).
Feel free to discuss on this here.
49 replies
ZetaX
Feb 27, 2007
NoDealsHere
May 4, 2019
2-var inequality
sqing   0
14 minutes ago
Source: Own
Let $ a,b > 0 ,   a+b+a^2+ b^2 +ab  = 3.$ Prove that
$$  \frac{1}{a}+ \frac{ 1}{ b }-\frac{6}{ 5ab } \geq\frac{2(\sqrt{10}-2)}{9}$$$$\frac{1}{a}+\frac{1}{b}-\frac{1}{ab } \geq\frac{4\sqrt{10}-5}{9}$$$$\frac{1}{a}+\frac{1}{b}+ \frac{1}{ab } \geq\frac{17+8\sqrt{10}}{9}$$
0 replies
sqing
14 minutes ago
0 replies
Rational Points in n-Dimensional Space
steven_zhang123   0
37 minutes ago
Let \( T = (x_1, x_2, \ldots, x_n) \), where \( x_i \) is rational for \( i = 1, 2, \ldots, n \). A vector \( T \) is called a rational point in \( n \)-dimensional space. Denote the set of all such vectors \( T \) as \( S \). For \( A = (x_1, x_2, \ldots, x_n) \) and \( B = (y_1, y_2, \ldots, y_n) \) in \( S \), define the distance between points \( A \) and \( B \) as \( d(A, B) = \sqrt{(x_1 - y_1)^2 + (x_2 - y_2)^2 + \cdots + (x_n - y_n)^2} \). We say that point \( A \) can move to point \( B \) if and only if there is a unit distance between two points in \( S \).

Prove:
(1) If \( n \leq 4 \), there exists a point that cannot be reached from the origin via a finite number of moves.
(2) If \( n \geq 5 \), any point in \( S \) can be reached from any other point via moves.
0 replies
steven_zhang123
37 minutes ago
0 replies
Inspired by old results
sqing   5
N 43 minutes ago by sqing
Source: Own
Let $a,b,c $ be reals such that $a^2+b^2+c^2=3$ .Prove that
$$(1-a)(k-b)(1-c)+abc\ge -k$$Where $ k\geq 1.$
$$(1-a)(1-b)(1-c)+abc\ge -1$$$$(1-a)(1-b)(1-c)-abc\ge -\frac{1}{2}-\sqrt 2$$
5 replies
sqing
Yesterday at 7:36 AM
sqing
43 minutes ago
equation in integers
Pirkuliyev Rovsen   2
N an hour ago by ytChen
Solve in $Z$ the equation $a^2+b=b^{2022}$
2 replies
+1 w
Pirkuliyev Rovsen
Feb 10, 2025
ytChen
an hour ago
Inequality for Sequences
steven_zhang123   0
2 hours ago
Given a positive number \( t \) and integers \( m, n \geq 2 \), prove that for any \( n \) positive numbers \( a_1, a_2, \ldots, a_n \) satisfying \( a_j - a_{j-1} \leq t \) for \( j = 1, 2, \ldots, n \) (with the convention \( a_0 = 0 \)), the following inequality holds:
\[
\sum_{j=1}^n a_j^{2m-1} \leq \frac{mt}{2} \left( \sum_{j=1}^n a_j^{m-1} \right)^2.
\]
0 replies
steven_zhang123
2 hours ago
0 replies
Circumcircle of ADM
v_Enhance   70
N 2 hours ago by Shan3t
Source: USA TSTST 2012, Problem 7
Triangle $ABC$ is inscribed in circle $\Omega$. The interior angle bisector of angle $A$ intersects side $BC$ and $\Omega$ at $D$ and $L$ (other than $A$), respectively. Let $M$ be the midpoint of side $BC$. The circumcircle of triangle $ADM$ intersects sides $AB$ and $AC$ again at $Q$ and $P$ (other than $A$), respectively. Let $N$ be the midpoint of segment $PQ$, and let $H$ be the foot of the perpendicular from $L$ to line $ND$. Prove that line $ML$ is tangent to the circumcircle of triangle $HMN$.
70 replies
v_Enhance
Jul 19, 2012
Shan3t
2 hours ago
Prove this relation in triangle ABC
Entrepreneur   1
N 3 hours ago by MathIQ.
Source: Conjectured by me
In $\Delta ABC,$ prove that$$\color{blue}{2\angle A=3\angle B\implies c^3a^3(c+a)^2=b^2(c^2+a^2-b^2+2ca)(c^2+a^2-b^2)^2.}$$
1 reply
Entrepreneur
Aug 1, 2024
MathIQ.
3 hours ago
45 degrees everywhere
Rijul saini   12
N 3 hours ago by guptaamitu1
Source: India IMOTC 2024 Day 2 Problem 2
Let $ABC$ be an acute angled triangle with $AC>AB$ and incircle $\omega$. Let $\omega$ touch the sides $BC, CA,$ and $AB$ at $D, E,$ and $F$ respectively. Let $X$ and $Y$ be points outside $\triangle ABC$ satisfying \[\angle BDX = \angle XEA = \angle YDC = \angle AFY = 45^{\circ}.\]Prove that the circumcircles of $\triangle AXY, \triangle AEF$ and $\triangle ABC$ meet at a point $Z\ne A$.

Proposed by Atul Shatavart Nadig and Shantanu Nene
12 replies
Rijul saini
May 31, 2024
guptaamitu1
3 hours ago
equation 2025
mohamed-adam   1
N 3 hours ago by MathIQ.
Source: own
Find all positive integers $a,b$ such that $$a^8-2b^5=2025b$$
1 reply
mohamed-adam
Yesterday at 7:49 PM
MathIQ.
3 hours ago
Graph Theory
ABCD1728   1
N 3 hours ago by ABCD1728
Can anyone provide the PDF version of "Graphs: an introduction" by Radu Bumbacea (XYZ press), thanks!
1 reply
ABCD1728
Yesterday at 5:10 AM
ABCD1728
3 hours ago
Rectangles of grid cells
tapir1729   11
N 3 hours ago by Mathandski
Source: TSTST 2024, problem 9
Let $n \ge 2$ be a fixed integer. The cells of an $n \times n$ table are filled with the integers from $1$ to $n^2$ with each number appearing exactly once. Let $N$ be the number of unordered quadruples of cells on this board which form an axis-aligned rectangle, with the two smaller integers being on opposite vertices of this rectangle. Find the largest possible value of $N$.

Anonymous
11 replies
tapir1729
Jun 24, 2024
Mathandski
3 hours ago
Interesting problem from a friend
v4913   11
N 4 hours ago by YaoAOPS
Source: I'm not sure...
Let the incircle $(I)$ of $\triangle{ABC}$ touch $BC$ at $D$, $ID \cap (I) = K$, let $\ell$ denote the line tangent to $(I)$ through $K$. Define $E, F \in \ell$ such that $\angle{EIF} = 90^{\circ}, EI, FI \cap (AEF) = E', F'$. Prove that the circumcenter $O$ of $\triangle{ABC}$ lies on $E'F'$.
11 replies
v4913
Nov 25, 2023
YaoAOPS
4 hours ago
Curious inequality
produit   2
N 4 hours ago by RainbowNeos
Positive real numbers x_1, x_2, . . . x_n satisfy x_1 + x_2 + . . . + x_n = 1.
Prove that
1/(1 −√x_1)+1/(1 −√x_2)+ . . . +1/(1 −√x_n)⩾ n + 4.
2 replies
produit
May 10, 2025
RainbowNeos
4 hours ago
Inspired by an old problem
RainbowNeos   0
4 hours ago
Given $n\geq 3$ and $a_i \geq 0, i=1,2,...,n$. Show that
\[\sum_{i=1}^n (a_i-a_{i+1})^3\leq \frac{1}{2\sqrt{3}} (\sum_{i=1}^n a_i )^3,\]where $a_{n+1}=a_1$.
0 replies
RainbowNeos
4 hours ago
0 replies
Perfect square preserving polynomial
Omid Hatami   35
N Apr 27, 2025 by joshualiu315
Source: Iran TST 2008
Find all polynomials $ p$ of one variable with integer coefficients such that if $ a$ and $ b$ are natural numbers such that $ a + b$ is a perfect square, then $ p\left(a\right) + p\left(b\right)$ is also a perfect square.
35 replies
Omid Hatami
May 25, 2008
joshualiu315
Apr 27, 2025
Perfect square preserving polynomial
G H J
Source: Iran TST 2008
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Omid Hatami
1275 posts
#1 • 4 Y
Y by anantmudgal09, bumjoooon, Adventure10, PikaPika999
Find all polynomials $ p$ of one variable with integer coefficients such that if $ a$ and $ b$ are natural numbers such that $ a + b$ is a perfect square, then $ p\left(a\right) + p\left(b\right)$ is also a perfect square.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Albanian Eagle
1693 posts
#2 • 3 Y
Y by Adventure10, Mango247, PikaPika999
very very nice problem. I don't have an elementary proof as of now so I would really like to see you solution.
I excluded P from having degree $ \geq 3$ so i'm assuming the only solutions are $ P(x)=k^2x$
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
gagan_goku
18 posts
#3 • 2 Y
Y by Adventure10, PikaPika999
can we say that if a polynomial P(x) takes square values for all integers , then
P(x) = {f(x)}^2
where f(x) = polynomial with integer coefficients
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
test20
988 posts
#4 • 3 Y
Y by Adventure10, PikaPika999, and 1 other user
gagan_goku wrote:
can we say that if a polynomial P(x) takes square values for all integers , then
P(x) = {f(x)}^2
where f(x) = polynomial with integer coefficients

Yes, that's a well-known theorem.

See for instance
H. Davenport, D.J. Lewis, A. Schinzel:
Polynomials of certain special types.
Acta Arithmetica 9, 1964, 107-116.
This post has been edited 1 time. Last edited by test20, May 26, 2008, 11:11 AM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
gagan_goku
18 posts
#5 • 2 Y
Y by Adventure10, PikaPika999
if thats true , then you can say that f(x,y) = P(x) + P(y^2 -x) = g(x,y) ^ 2
maybe this yields something
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
fearailhold
24 posts
#6 • 2 Y
Y by Adventure10, PikaPika999
I've been spending a lot of time mulling this one over in my head. Can anyone solve it?
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Albanian Eagle
1693 posts
#7 • 3 Y
Y by Adventure10, Mango247, PikaPika999
If $ P$ is not an odd polynomial there exists an $ a$ so that $ P(x)+P(a)$ is not divisible by $ x+a$
$ P(x^2-a)+P(a)=Q(x)^2$ so $ P(x)+P(a)=Q(\sqrt{x+a})^2$ and $ P(x)+P(a)=R(x)^2$ also $ P(x)+P(b)=K(x)^2$
but $ R(x)^2-K(x)^2=c$ can not hold as an identity so $ c=0$ and the polynomial is a constant function $ P(x)=2m^2$
if P is an odd polynomial $ P(x)+P(a)=(x+a)R(x)^2$ and $ P(x)+P(b)=(x+b)K(x)^2$. how can we show that these last two equations imply some contradiction if the degree of $ P$ is too large?...
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
fearailhold
24 posts
#8 • 3 Y
Y by Adventure10, Mango247, PikaPika999
Albanian Eagle wrote:
If $ P$ is not an odd polynomial there exists an $ a$ so that $ P(x) + P(a)$ is not divisible by $ x + a$

Do you have a proof of this?
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
t0rajir0u
12167 posts
#9 • 3 Y
Y by Adventure10, Mango247, PikaPika999
Albanian Eagle wrote:
i'm assuming the only solutions are $ P(x) = k^2x$

Of course there's also $ P(x) = 2k^2$.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ali666
352 posts
#10 • 3 Y
Y by Adventure10, Mango247, PikaPika999
Albanian Eagle wrote:
and $ P(x) + P(a) = R(x)^2$
why?...can you say more details please?
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
harazi
5526 posts
#11 • 3 Y
Y by Adventure10, Mango247, PikaPika999
I will assume the answers to the preceeding questions have already been given, so let me try to answer Albanian Eagle's question. Suppose that $ (X+a)R(x)^2-(X+b)P(X)^2=c$, a nonzero constant (otherwise the conclusion is obvious...). Then obviously $ P$ and $ R$ are relatively prime and have the same degree. Differentiate the relation to obtain $ (X+a)2RR'+R^2-P^2-(X+b)2PP'=0$, thus $ R$ divides $ P+2(X-b)P'$ and thus must be a constant times this thing, by the remark on the degree. This gives you a differential equation on $ P$ and allows finding all solutions, but unfortunately it may very well happen that there are solutions of arbitrarily large degree (and it happens, actually...), but this imposes however very strong conditions on $ P$. Now, by choosing some other value $ d,e,..$ for $ a,b$, I think we can conclude. I'm still puzzled by this apparently extremely easy problem, which however does not seem to have a decent solution.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Albanian Eagle
1693 posts
#12 • 2 Y
Y by Adventure10, PikaPika999
Oh nice..
so Harazi can substitute "a constant times" with "equal" by considering leading coefficients.
and so we have $ R=K+2(x+b)K'$ but also vice versa so
$ (x+b)K'+(x+a)R'=0$ and particularly
$ K'=R'=...$ by putting different choices So $ P(x)+P(m)=(x+m)(Q(x)+c_m)^2$ holds for some fixed $ Q$
this is possible only when $ Q$ is constant so $ P$ is linear and we can easily conclude
:lol: we don't even need the differential equations.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
harazi
5526 posts
#13 • 2 Y
Y by Adventure10, PikaPika999
But I love differential equations! Just kidding, I really hate them, thanks for pointing out this simplification, I had no paper with me to check if indeed it can be made easier. Anyway, great problem.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ali666
352 posts
#14 • 2 Y
Y by Adventure10, PikaPika999
fearailhold wrote:
Albanian Eagle wrote:
If $ P$ is not an odd polynomial there exists an $ a$ so that $ P(x) + P(a)$ is not divisible by $ x + a$

Do you have a proof of this?
if for all $ a$, $ P(x)+P(a)$ is diviseble by $ x+a$ then $ G(x)=P(x)+P(-x)$ is zero for all $ a$....and P is odd...
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Albanian Eagle
1693 posts
#15 • 2 Y
Y by Adventure10, PikaPika999
:lol:
If the words Iran and Polynomial appear in the same thread it can never disappoint you :D
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
gagan_goku
18 posts
#16 • 3 Y
Y by Adventure10, Mango247, PikaPika999
1 doubt
how does $ P(x) + P(a) = P(\sqrt{x+a})^2$ imply $ P(x) + P(a) = R(x)^2$
secondly , if you have $ P(x) + P(a) = R(x,a)^2$ , then i think we can conclude that the degree of $ P$ is 0
but then we dont get the solution $ P(x) = k^2x$
does this solution come from $ P$ being odd
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
rchlch
35 posts
#17 • 3 Y
Y by Adventure10, Mango247, PikaPika999
gagan_goku wrote:
can we say that if a polynomial P(x) takes square values for all integers , then
P(x) = {f(x)}^2
where f(x) = polynomial with integer coefficients
This claim should be slightly modified, cause $ f(x)$ do not need to be with integer coefficients. For example $ P(x)=\frac{x^2(x-1)^2}{4}$.
Then I think I have a elementary proof to this fact.
The most important case is that the degree of $ P$ is even. In this case we let $ \deg P=2n$ and tha cofficent of the $ 2n-th$ power is $ A$.
Consider $ L=\lim_{x \rightarrow \infty}{(\sqrt{P(x+n-1)} -C_{n-1}^1 \sqrt{P(x+n-2)}+\ldots+(-1)^{n-1} \sqrt{P(x)})}$
Here is the most important result:
We claim that $ L$ is a constant.
In fact, we can find a polynomial $ Q(x)$ such that $ \deg Q=n$ and $ \deg(P-AQ^2) \leq n-1$ (This is true due to Newton's formula for symmitric sums.)
Then we have
${ \lim_{x \rightarrow \infty}{\sqrt{P(x)}-\sqrt{A}Q(x)}=\lim_{x \rightarrow \infty}{\frac{P(x)-AQ^2(x)}{\sqrt{P(x)}+\sqrt{A}Q(x)}}}=0$

So
$ L=\sqrt{A} \lim_{x \rightarrow \infty}{(Q(x+n-1) -C_{n-1}^1 Q(x+n-2)+\ldots+(-1)^{n-1} Q(x))}=  \sqrt{A} (n-1)!$
is a constant.
Finally, suppose $ P(x)=a_x^2$ for all $ x \in Z$. Then $ \lim_{x \rightarrow \infty}{(a_{x+n-1}-C_{n-1}^1 a_{x+n-2}+ \ldots +(-1)^{n-1} a_{x})}=L$. Therefore $ L$ is a integer, and for large enough $ x \in Z$, we have ${ a_{x+n-1}-C_{n-1}^1 a_{x+n-2}+ \ldots +(-1)^{n-1} a_{n}}=L$. Note that the sequence $ {a_n}$ has the fomula $ a_n=f(n)$ where $ \deg{f}=n$ for large enough $ n$. So we conclude $ P(x)=f^2(x)$ for large enough $ x \in Z$ and therefore $ P(x)=f^2(x)$ for all $ x$. And it's obvious that $ f(x) \in Q[x]$.

In the case that the degree of $ P$ is odd, let $ P'(x)=P(x^2)$ we do the same thing, and it's easy to claim that such $ P(x)$ don't exist.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
test20
988 posts
#18 • 3 Y
Y by Adventure10, Mango247, PikaPika999
rchlch wrote:
gagan_goku wrote:
can we say that if a polynomial P(x) takes square values for all integers , then
P(x) = {f(x)}^2 where f(x) = polynomial with integer coefficients
This claim should be slightly modified, cause $ f(x)$ do not need to be with integer coefficients. For example $ P(x) = \frac {x^2(x - 1)^2}{4}$.

Aaah, yes indeed. The correct statement is:

If a MONIC polynomial $ P(x)$ takes square values for all integers $ x$,
then $ P(x) = {f(x)}^2$ where $ f(x)$ is a polynomial with integer coefficients.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Albanian Eagle
1693 posts
#19 • 2 Y
Y by Adventure10, PikaPika999
There are stronger lemmas out there
like http://www.mathlinks.ro/Forum/viewtopic.php?search_id=1576618391&t=148696
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
rchlch
35 posts
#20 • 3 Y
Y by Adventure10, Mango247, PikaPika999
Albanian Eagle wrote:
I think this lemma cannot be used in this problem because no direct evidence shows that $ P(x)$ is a monic polynomial.
And for your lemma, one of my classmates has given a beautiful solution.
Thank you all the same.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Albanian Eagle
1693 posts
#21 • 3 Y
Y by Adventure10, Mango247, PikaPika999
rchlch wrote:
Albanian Eagle wrote:
I think this lemma cannot be used in this problem because no direct evidence shows that $ P(x)$ is a monic polynomial.
And for your lemma, one of my classmates has given a beautiful solution.
Thank you all the same.
It was just a reply to test20's post :lol:
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ith_power
233 posts
#22 • 6 Y
Y by toto1234567890, Adventure10, Mango247, Mango247, Mango247, PikaPika999
My solution was too big(6 pages). however here is an outline:
Click to reveal hidden text
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
amir.sepehri
3 posts
#23 • 4 Y
Y by ali.agh, Adventure10, Mango247, PikaPika999
I give this exam and i couldn't solve this problem :(
in fact only one of students solved this problem, his main idea is : if a+b is square then a*x^2+b*x^2 is square too, so
p(ax^2)+p(bx^2) is perfect square for all integer x so there exist a g(x) such: p(ax^2)+p(bx^2)=g(x)^2
if p(x)=(a_n)*x^n+.... then (a_n)*(a^n+b^n)=u^2 (g(x)=u*x^k+...) for all a,b (a+b=perfect square) let a=b=2 so
(a_n)*2^(n+1)=u^2 and ... it's easy to see that p(x)=(k^2)*(x^2) or 2*(k^2), so we solve this hard problem
:wink:
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
shekast-istadegi
81 posts
#24 • 2 Y
Y by Adventure10, PikaPika999
Theorem_ let $f$ be a integer polynomial.We have $f(a+tp^s)\equiv {}^{p{}^{s+1}}f(a)+tp^sf{}'(a)$such that $(t,p)=1$.So by above theorem we have for any integer polynomial $Q$ which has not any double root ,there are infinitely many prime $p$ which there exist some $n$ such that $p\mid Q(n)$ but $ Q(n)\not\equiv {}^p{}^{2}0$.now suppose $p(a)+p(x^2-a)=Q(x)$ .(step 1)if $Q$ has not any double root ,we have $p\mid Q(n)$ but $ Q(n)\not\equiv {}^p{}^{2}0$ a contraction.(step 2)if $Q$ has some double root, we have $Q(x)=R_{a}(x)T_{a}(x)^2$ such that $R_{a}(x)$ has not any double root .$Q(m)$ is a perfect square for any positive integer $m$ so $R_{a}(Y)$ is a perfect square for all $y> M$ and we have a contraction same way to step 1.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
toto1234567890
889 posts
#25 • 3 Y
Y by Adventure10, Mango247, PikaPika999
And how to solve the main problem?
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
anantmudgal09
1980 posts
#26 • 13 Y
Y by rafayaashary1, v_Enhance, llpllp, Ankoganit, bumjoooon, mijail, Supercali, ywq233, centslordm, guptaamitu1, Adventure10, math_comb01, PikaPika999
Solution

Credits and remarks
This post has been edited 1 time. Last edited by anantmudgal09, Sep 23, 2016, 4:44 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ywq233
38 posts
#28 • 2 Y
Y by Adventure10, PikaPika999
anantmudgal09 wrote:
and if $d$ is odd then we have $$\left((t^2-1)^{\frac{d-1}{2}}-1\right)^2<\left(\frac{g(t)}{tc}\right)^2<\left((t^2-1)^{\frac{d-1}{2}}\right)^2$$which is again a contradiction! Thus, we must have $d=1$.
How can you get the left side?
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
bumjoooon
165 posts
#29 • 4 Y
Y by Mango247, Mango247, Mango247, PikaPika999
anantmudgal09 wrote:
Lemma 2. Let $f$ be a non constant polynomial with integer such that $\sqrt{f(n)}$ is an integer for all positive integers $n$. Then there exists a polynomial $g(x)$ with integer coefficients such that $f(x)=g(x)^2$.

I learned another simple way to prove this lemma from lectures on winter camp in Korea around 2018.

Proof) Think of $f(x+m)f(x)$ for some natural number $m$. If $m$ is sufficiently large, we can separate complex roots of $f(x)$ and $f(x+m)$ and therefore $f(x+m)$ and $f(x)$ is relatively prime. By assumption, $f(x+m)f(x)$ is even degree and the maximal coefficient is square. Therefore, $f(x+m)f(x)=h(x)^2$ for some $h(x) \in \mathbb{Z}[x]$. Since $f(x)$ and $f(x+m)$ is prime, you can easily see that $f(x)=g(x)^2$ for some $g(x) \in \mathbb{Z}[x]$.

Same idea is used in Proposition mentioned in https://artofproblemsolving.com/community/u314211h1985879p14247495
This post has been edited 2 times. Last edited by bumjoooon, Mar 10, 2020, 1:39 AM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
bumjoooon
165 posts
#30 • 1 Y
Y by PikaPika999
Another astonishing proof for Lemma by my student.

Proof) Let $f(x)=a_n x^n +....+a_0$. By assumption, $a_n >0$.
[Case 1] $n$ is odd.
Multiply the term $(a_n x +1)$. There are infinitely many $x$ such that $a_n x+1$ is perfect square. Therefore, infinitely many $x$ such that $(a_n x+1)f(x)$ is perfect square. However, there are also $x$ such that $a_n x +1$ is not a perfect square and contradiction.
[Case 2] $n$ is even, but $a_n$ is not a perfect square.
Multiply the term $(a_n x^2 +1)$. And same as above. Here we used the fact that pell equation $y^2 - a_n x^2 =1$ has infinitely many solution.

Therefore, $n$ is even and $a_n$ is perfect square and $f(x)=g(x)^2$ for some $g(x) \in \mathbb{Z}[x]$.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
primesarespecial
364 posts
#31 • 2 Y
Y by PRMOisTheHardestExam, PikaPika999
Here is a smol bren solution.
Write the polynomial,as $x^d.P(x)$, with $P(0)$ non zero.Assume $P$ non-constant.Then assuming,$P$ is not a square ,write it as a square times a bunch of irreducibles.Consider such a irreducible ,say $h(x)$,and then by Schur ,choose a very large prime factor of the polynomial $h(x^2)$.Then using Bezout,we can conclude that if $p|h(m^2)$,then $p$ does not divide any other irreducible factor of $P$ evaluated at $m^2$.Then,again using bezout but this time with $h'(x)$,we conclude that if $p$,is large enough ,then $p$ also doesn't divide $h'(m^2)$,thus from Hensel,we can uniquely lift the root corresponding to $m^2$ mod $p$ to mod $p^2$ .Due to this unique lift ,we can find such an $n$,such that $n \equiv m^2$ mod $p$ but $p^2$ does not divide $P(n)$.Thus we have that $v_p(P(n))$ is odd.Now choose a $b$ such that $v_p(b)=j$ is even and quite large.As $p$ is large (forgot to mention $p$ doesn't divide $P(0)$,so $p$ doesn't divide $P(b)$).
Now,we use hensel's again.Firstly note that $n$ is a quadratic residue mod $p$.So by hensel's on the polynomial $x^2-n$,we see that the derivative is non zero mod $p$.Thus,we can find a solution to this poly mod $p^{j+1}$,such that $v_p(x^2-n)$ is exactly $j$.Thus we have found $b+n$ is a perfect square but $v_p(n^d.P(n)+b^d.P(b))$ is odd which is a contradiction.Thus,$P(x)$ is a perfect square.

Using a very similar argument as above,we can prove that $d$ is even(Basically choosing $a,b$,such that $P(a)$ has a large $v_p$,which is obv
even,except here we choose $b$ to have small odd $v_p$,and obv $p$ doesn't divide $P(b)$.$a+b$ being a perfect square is ensured by the very similar hensel argument as above).
So we have that we can write $P(a)+P(b)$ as $a^{2d}Q(a)^2+b^{2d}Q(b)^2$.So if we fix $a$,we see that $a^dQ(a)$ is a part of infinitely many Pythagorean triples,which is impossible.
Observing the left out cases(which are not that hard),we get our solutions.
This post has been edited 2 times. Last edited by primesarespecial, Sep 2, 2022, 12:59 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
VicKmath7
1391 posts
#32 • 1 Y
Y by PikaPika999
As in #26, $f(ax^2)+f(bx^2)$ is always a square for a fixed pair $(a,b)$ with $a+b$ being a square, so $g(x)^2=f(ax^2)+f(bx^2)$ for some integer polynomial $g(x)$. In addition, we consider only the case $\deg(f) \geq 2$.

Next, we consider the leading coefficient of this polynomial; it is of the form $T(a^n+b^n)$ ($T$ is the leading coefficient of $f$), which shall be a square. If $n$ is even, plug $(1,p^2-1),(1,p^2+2p)$ for some prime $p$ to see that $2T, T$ are QR's modulo $p$ for all primes $p$, so $2$ is a QR for all primes $p$, absurd, since $(\frac{2}{p})=(-1)^{\frac{p^2-1}{8}}$. Now we can finish the other case with $(2,2),(3,1)$ : $2^{n+1}T$ is a square, hence $T$ is a square, so $3^n+1=x^2$ is a square, so $3^n=(x-1)(x+1)$ and thus $x-1=1$ and $n=1$, contradiction.
This post has been edited 2 times. Last edited by VicKmath7, Sep 25, 2022, 2:47 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
giglio
9 posts
#33 • 1 Y
Y by PikaPika999
I see pleny of solutions over here depending on this theorem about a polynomial being square too many times. I solved to problem using a totally different approach with pells equations, some estimations of the growth of some terms, and an unexpected use of Norms function at the end. I will write down the solution tomorrow.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
giglio
9 posts
#34 • 3 Y
Y by jrsbr, Dissonant, PikaPika999
Suppose $degP > 0$. If $P$ is constant so $2P(2)$ must be a square, so $P \equiv 2k^2$ for some integer $k$. Indeed all of these work.
Also suppose that the leading coefficient of $P$ is positive. If it's negative then $P(2N^2)$ is negative for big $N$, but is must also be twice a perfect square, contradiction.
Fix some positive integer $t$ such that the equation $b^2 - 2a^2=t$ has a solution.
We know there are infinetely many big $t$'s satisfying this (perfect squares for instance).
By pell's equation we know that, given one solution $b_0^2 - 2a_0^2$, we can find infinetely many solutions of the form
$$b_n + a_n \sqrt2= (b_0 + a_0 \sqrt2)(3+2sqrt2)^n.$$Just so I'll write quicker, set $ r = 3 + 2 \sqrt2$, we'll use it at lot.
Now we know that, as $2a_n^2 + 2a_n^2$ is the square of an integer, $2P(a_n^2)$ must be the square of an even integer, say, by definition, $4r_n^2$. So $P(a_n^2) = 2r_n^2$.
We also know that $b_n^2 - t + t$ is a perfect square, so $P(b_n^2 - t) = s_n^2 - P(t)$, for some integer $s_n$. But, by definition, $b_n^2 - t = 2a_n^2$, so $2r_n^2$ and $s_n^2 - P(t)$ must be equal. So we have that $s_n^2 - 2r_n^2 = P(t)$.
But this means that $s_n, r_n$ are pell's solutions for $x^2 - 2y^2 = P(t)$. This means that $$s_n + r_n \sqrt2 = (s_0 + r_0 \sqrt2)r^{t_n}$$for some minimal solution $(s_0,r_0)$ and some integer $t_n$. As there are only finitely many minimal solutions for pell's equation, by the pigeon hole principle one of them must appear infinetely many times, varying $n$. WLOG call it $(s_0,r_0)$.
In other words, we can say that
$$s_n = \frac{s_0 + r_0 \sqrt2}{2}r^{t_n} + \frac{s_0 - r_0 \sqrt2}{2}r^{-t_n}$$so
$$s_n^2 = \big( \frac{s_0 + r_0 \sqrt2}{2} \big)^2 r^{2t_n} + \big( \frac{s_0 - r_0 \sqrt2}{2} \big)^2 r^{-2t_n} + 2\frac{s_0^2 - 2r_0^2}{4} = (1+o(1))\big( \frac{s_0 + r_0 \sqrt2}{2} \big)^2 r^{2t_n}$$So this is one estimate for $s_n^2$.
Let $d = degP$ and $c$ the leading coefficient. Now we also know that
$$s_n^2 = P(b_n^2 - t) + P(t) = (1+o(1))P(b_n^2) = (1+o(1))c.b_n^{2d}.$$As $$b_n = (1+o(1))\frac{b_0+a_0 \sqrt2}{2}r^n,$$$$s_n^2 = (1+o(1))c.\big( \frac{b_0+a_0 \sqrt2}{2} \big) ^{2d}r^{2nd}$$Therefore the ratio between the $2$ estimates converges to $1$. So
$$lim_{n \to \infty} \frac{\big( \frac{s_0 + r_0 \sqrt2}{2} \big)^2 r^{2t_n}}{c.\big( \frac{b_0+a_0 \sqrt2}{2} \big) ^{2d}r^{2nd}} = 1.$$So
$$lim_{n \to \infty} \frac{\big( \frac{s_0 + r_0 \sqrt2}{2} \big)^2}{c.\big( \frac{b_0+a_0 \sqrt2}{2} \big) ^{2d}} r^{2t_n - 2dn} = 1.$$So, a constant times an integer power of $r$ converges to $1$. Due to the fact the the sequence $C.r^m$ is discrete, if it converges to $1$, it's eventually $1$. So there exists an integer $q$ such that
$$\frac{\big( \frac{s_0 + r_0 \sqrt2}{2} \big)^2}{c.\big( \frac{b_0+a_0 \sqrt2}{2} \big) ^{2d}} r^q = 1.$$Rearranging things and substituting $r$ back to $3+2 \sqrt2$, we have
$$(s_0 + r_0 \sqrt2)^2 (3+2 \sqrt2)^t = c(b_0 + a_0 \sqrt2)^{2d} 2^{2d-2}$$Now we are gonna make use of the Norm function defined by $N(m + n \sqrt2) = m^2 - 2n^2$, where $m,n$ are rationals. One can easily prove that this function is well defined and is multiplicative. So we have
$$(s_0^2 - 2r_0^2)^2 . 1^t = c^2(b_0^2 - 2a_0^2)^2 2^{4d-4}$$So
$$P(t)^2 = c^2 . t^{2d} . 2^{4d-4}$$As the leading coefficient of $P$ is positive, $P$ is eventually positive, so we can take the squareroot of both sides to conclude that
$$P(t) = c.t^d.2^{2d-2}$$But notice that this is true for infinetely many big integers $t$'s, so it must be true that
$$P(x) = x^d.c.2^{2d-2}$$But remember that $c$ is the leading coefficient of $P$, so $c = c.2^{2d-2}$, so $2^{2d-2}=1$, so $d$ must be $1$, and $P(x) = cx$.
Now, as $2P(2)$ is a square, so $4c$ is a square, so $c$ itself is a perfect square. Given that, clearly $a+b$ is a square if and only if ac+bc is a square.
Therefore all solutions are $P(x) = k^2x$ and $ P(x) = 2k^2$, where $k$ is an arbitrary integer.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Pyramix
419 posts
#35 • 1 Y
Y by PikaPika999
We use a few well-known theorems to solve.

Theorem 1.
A polynomial is a perfect square at every integer iff it is a perfect square in $\mathbb{Z}[x]$

Theorem 2. (Mihailescu)
For $a,b,c,d>1$ being positive integers, the Diophantine $a^b-c^d=1$ has a unique solution, which is $(a,b,c,d)=(3,2,2,3)$.

For any $x$, we have $2P(2x^2)$ is a perfect square and $P(x^2)+P(3x^2)$ is also a perfect square.

So, from theorem 1 there are polynomials $A(x),B(x)\in\mathbb{Z}[x]$ such that
\[2P(2x^2)=A(x)^2,\ \ P(x^2)+P(3x^2)=B(x)^2\]Let $k$ be the leading coefficient of $P$ and $d=\deg P>0$.
Then we have $2^{d+1}k=u^2$ and $(3^d+1)k=v^2$ for some integers $u,v$.

If $d$ is even then $k=2w^2$ for some $w$, which means $3^d+1=2y^2$ for some $y$, which is a contradiction $\pmod{3}$.
So, $d$ is odd. But then $k$ is a perfect square from first equation, which means $3^d+1$ is a perfect square. From theorem 2, $d=1$ is forced.

Finally, note that for $d=1$ only $P(x)=c^2x$ works for some $c$ and for $d=0$ only $P(x)=2c^2$ works. These are the only solutions.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
HamstPan38825
8868 posts
#36 • 1 Y
Y by PikaPika999
The key lemma is the following:

Lemma: Let $P(x)$ be a polynomial such that $P(n)$ is a perfect square for all sufficiently large positive integers $n \geq 1$. Then there exists a polynomial $Q(x)$ such that $P(x) = Q(x)^2$ identically.

Proof: This is a special case of Steps 3 and 4 to this proof to USEMO 2019/2.

Now we just kind of destroy the problem. Observe that for each fixed positive integer $k$, we have the numerical equality \[P\left(n^2 - k\right) + P(k) = P_k(n)^2\]for some function $P_k(n)$ for all sufficiently large $n$. Thus the polynomial $P\left(n^2-k\right) + P(k)$ in $n$ takes only perfect square values, and by the lemma, it follows that it must be identically equal to $P_k(n)^2$ for some polynomial $P_k$.

On the other hand, the expression on the LHS is even, hence $P_k(n)$ is either even or odd. (Otherwise, it is easy to construct an $x$ such that $|P_k(x)| \neq |P_k(-x)|$.) We make the following claim:

Claim: Either $P_k(x)$ is even for all $k \geq 1$ or odd for all $k \geq 1$.

Proof: Suppose there existed a $k$ such that $P_k(x)$ was odd and $P_{k+1}$ was even. Then we can write
\begin{align*}
P\left(x^2-k\right) + P(k) = x^2 Q_k\left(x^2\right)^2 &\implies P(x-k) + P(k) = xQ_k(x)^2 \\
P\left(x^2-k-1\right) + P(k+1) = Q_{k+1}\left(x^2\right)^2 &\implies P(x-k-1) + P(k+1) = Q_{k+1}(x)^2
\end{align*}for some polynomials $Q_k(x)$ and $Q_{k+1}(x)$. So in particular, we have the equality \[Q_{k+1}(x+1)^2 - P(k+1) = P(x-k) = xQ_k(x)^2 - P(k).\]However, taking $x$ to be a sufficiently large negative number yields a negative RHS and positive LHS, hence contradiction. The symmetric case works out analogously. $\blacksquare$

So now we have to resolve two cases.

First Case: Suppose that $P_k(x)$ is even for all $x$. Then the same equation in the previous display reads \[Q_{k+1}(x+1)^2 - P(k+1) = Q_k(x)^2 - P(k).\]So for $Q_{k+1}'(x) = Q_{k+1}(x+1)$ and $Q_k(x)$, the coefficients of the squares of these polynomials agree everywhere except the constant term. It follows easily that their constant terms must also agree, thus $P(k) = P(k+1)$ for all positive integers $k$, and $P$ is constant. We can verify $P \equiv 2c^2$ in this case.

Second Case: Suppose that $P_k(x)$ is odd for all $x$. Then by setting $x=0$ in the polynomial equality \[P(x-k) + P(k) = xQ_1(x)^2\]for all positive integers $k \geq 1$, we get that $P(k) + P(-k) = 0$, i.e. $P$ is odd. Then setting $x=1$ in this sequence of equalities, note that $P(k)-P(k-1)= Q_k(1)^2$ is always a perfect square. So reading as a polynomial, $P(k) - P(k-1) = R(k)^2$ is the perfect square of a polynomial. By applying the same argument with $x=4$, we obtain that $R(x)^2+R(x+1)^2+R(x+2)^2+R(x+3)^2 = S(x)^2$ for some polynomial $S(x)$, as the numerical equality holds for all sufficiently large $x$. [EDIT: the rest of this solution ended up being a fakesolve, so I would greatly appreciate anyone who could finish it off from here.]

Remark: This problem is a nightmare.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
joshualiu315
2534 posts
#37 • 1 Y
Y by PikaPika999
The answer is $P(x) = \boxed{2c^2, c^2x}$, which can be manually confirmed to be the only valid polynomials with degree less than $2$. Let $P(x) = a_0+a_1x+\dots+a_dx^d$ so that it has degree $d$.

Notice that plugging in $(a,b) = (2x^2, 2x^2)$ and $(a,b) = (x^2, 3x^2)$ gives

\[2P(2x^2) = 2a_0+4a_1x^2+\dots+2^{d+1}a_dx^{2d} = [Q(x)]^2\]\[P(x^2)+P(3x^2) = 2a_0+4a_1x^2+\dots+[1+3^d]a_dx^{2d} = [R(x)]^2,\]
where $Q, R \in \mathbb{Z}[x]$ and both have degree $d$. Let the leading coefficients of $Q(x)$ and $R(x)$ be $m$ and $n$, respectively. Then,

\[m^2 = 2^{d+1}a_d\]\[n^2 = a_d(1+3^d),\]
where $m$ and $n$ are positive integers.

If $d$ is even, then $a_d = 2c^2$ for some integer $c$, so

\[\left(\frac{n}{c} \right)^2 = 2(1+3^d) \equiv 2 \pmod{3},\]
when $d \neq 0$. Thus, there are no solutions for even $d$ except for $d=0$.

If $d$ is odd, then $a_d = c^2$ for some integer $c$, so

\[\left(\frac{n}{c} \right)^2 - 3^d = 1,\]
which by Mihailescu only has $d=1$ as a solution. Hence, there are no solutions when $d > 1$.
Z K Y
N Quick Reply
G
H
=
a