Plan ahead for the next school year. Schedule your class today!

Contests & Programs AMC and other contests, summer programs, etc.
AMC and other contests, summer programs, etc.
3 M G
BBookmark  VNew Topic kLocked
Contests & Programs AMC and other contests, summer programs, etc.
AMC and other contests, summer programs, etc.
3 M G
BBookmark  VNew Topic kLocked
G
Topic
First Poster
Last Poster
k a July Highlights and 2025 AoPS Online Class Information
jwelsh   0
Jul 1, 2025
We are halfway through summer, so be sure to carve out some time to keep your skills sharp and explore challenging topics at AoPS Online and our AoPS Academies (including the Virtual Campus)!

[list][*]Over 60 summer classes are starting at the Virtual Campus on July 7th - check out the math and language arts options for middle through high school levels.
[*]At AoPS Online, we have accelerated sections where you can complete a course in half the time by meeting twice/week instead of once/week, starting on July 8th:
[list][*]MATHCOUNTS/AMC 8 Basics
[*]MATHCOUNTS/AMC 8 Advanced
[*]AMC Problem Series[/list]
[*]Plus, AoPS Online has a special seminar July 14 - 17 that is outside the standard fare: Paradoxes and Infinity
[*]We are expanding our in-person AoPS Academy locations - are you looking for a strong community of problem solvers, exemplary instruction, and math and language arts options? Look to see if we have a location near you and enroll in summer camps or academic year classes today! New locations include campuses in California, Georgia, New York, Illinois, and Oregon and more coming soon![/list]

MOP (Math Olympiad Summer Program) just ended and the IMO (International Mathematical Olympiad) is right around the corner! This year’s IMO will be held in Australia, July 10th - 20th. Congratulations to all the MOP students for reaching this incredible level and best of luck to all selected to represent their countries at this year’s IMO! Did you know that, in the last 10 years, 59 USA International Math Olympiad team members have medaled and have taken over 360 AoPS Online courses. Take advantage of our Worldwide Online Olympiad Training (WOOT) courses
and train with the best! Please note that early bird pricing ends August 19th!
Are you tired of the heat and thinking about Fall? You can plan your Fall schedule now with classes at either AoPS Online, AoPS Academy Virtual Campus, or one of our AoPS Academies around the US.

Our full course list for upcoming classes is below:
All classes start 7:30pm ET/4:30pm PT unless otherwise noted.

Introductory: Grades 5-10

Prealgebra 1 Self-Paced

Prealgebra 1
Wednesday, Jul 16 - Oct 29
Sunday, Aug 17 - Dec 14
Tuesday, Aug 26 - Dec 16
Friday, Sep 5 - Jan 16
Monday, Sep 8 - Jan 12
Tuesday, Sep 16 - Jan 20 (4:30 - 5:45 pm ET/1:30 - 2:45 pm PT)
Sunday, Sep 21 - Jan 25
Thursday, Sep 25 - Jan 29
Wednesday, Oct 22 - Feb 25
Tuesday, Nov 4 - Mar 10
Friday, Dec 12 - Apr 10

Prealgebra 2 Self-Paced

Prealgebra 2
Friday, Jul 25 - Nov 21
Sunday, Aug 17 - Dec 14
Tuesday, Sep 9 - Jan 13
Thursday, Sep 25 - Jan 29
Sunday, Oct 19 - Feb 22
Monday, Oct 27 - Mar 2
Wednesday, Nov 12 - Mar 18

Introduction to Algebra A Self-Paced

Introduction to Algebra A
Tuesday, Jul 15 - Oct 28
Sunday, Aug 17 - Dec 14
Wednesday, Aug 27 - Dec 17
Friday, Sep 5 - Jan 16
Thursday, Sep 11 - Jan 15
Sunday, Sep 28 - Feb 1
Monday, Oct 6 - Feb 9
Tuesday, Oct 21 - Feb 24
Sunday, Nov 9 - Mar 15
Friday, Dec 5 - Apr 3

Introduction to Counting & Probability Self-Paced

Introduction to Counting & Probability
Wednesday, Jul 2 - Sep 17
Sunday, Jul 27 - Oct 19
Monday, Aug 11 - Nov 3
Wednesday, Sep 3 - Nov 19
Sunday, Sep 21 - Dec 14 (1:00 - 2:30 pm ET/10:00 - 11:30 am PT)
Friday, Oct 3 - Jan 16
Sunday, Oct 19 - Jan 25
Tuesday, Nov 4 - Feb 10
Sunday, Dec 7 - Mar 8

Introduction to Number Theory
Tuesday, Jul 15 - Sep 30
Wednesday, Aug 13 - Oct 29
Friday, Sep 12 - Dec 12
Sunday, Oct 26 - Feb 1
Monday, Dec 1 - Mar 2

Introduction to Algebra B Self-Paced

Introduction to Algebra B
Friday, Jul 18 - Nov 14
Thursday, Aug 7 - Nov 20
Monday, Aug 18 - Dec 15
Sunday, Sep 7 - Jan 11
Thursday, Sep 11 - Jan 15
Wednesday, Sep 24 - Jan 28
Sunday, Oct 26 - Mar 1
Tuesday, Nov 4 - Mar 10
Monday, Dec 1 - Mar 30

Introduction to Geometry
Monday, Jul 14 - Jan 19
Wednesday, Aug 13 - Feb 11
Tuesday, Aug 26 - Feb 24
Sunday, Sep 7 - Mar 8
Thursday, Sep 11 - Mar 12
Wednesday, Sep 24 - Mar 25
Sunday, Oct 26 - Apr 26
Monday, Nov 3 - May 4
Friday, Dec 5 - May 29

Paradoxes and Infinity
Mon, Tue, Wed, & Thurs, Jul 14 - Jul 16 (meets every day of the week!)
Sat & Sun, Sep 13 - Sep 14 (1:00 - 4:00 PM PT/4:00 - 7:00 PM ET)

Intermediate: Grades 8-12

Intermediate Algebra
Sunday, Jul 13 - Jan 18
Thursday, Jul 24 - Jan 22
Friday, Aug 8 - Feb 20
Tuesday, Aug 26 - Feb 24
Sunday, Sep 28 - Mar 29
Wednesday, Oct 8 - Mar 8
Sunday, Nov 16 - May 17
Thursday, Dec 11 - Jun 4

Intermediate Counting & Probability
Sunday, Sep 28 - Feb 15
Tuesday, Nov 4 - Mar 24

Intermediate Number Theory
Wednesday, Sep 24 - Dec 17

Precalculus
Wednesday, Aug 6 - Jan 21
Tuesday, Sep 9 - Feb 24
Sunday, Sep 21 - Mar 8
Monday, Oct 20 - Apr 6
Sunday, Dec 14 - May 31

Advanced: Grades 9-12

Calculus
Sunday, Sep 7 - Mar 15
Wednesday, Sep 24 - Apr 1
Friday, Nov 14 - May 22

Contest Preparation: Grades 6-12

MATHCOUNTS/AMC 8 Basics
Tues & Thurs, Jul 8 - Aug 14 (meets twice a week!)
Sunday, Aug 17 - Nov 9
Wednesday, Sep 3 - Nov 19
Tuesday, Sep 16 - Dec 9
Sunday, Sep 21 - Dec 14 (1:00 - 2:30 pm ET/10:00 - 11:30 am PT)
Monday, Oct 6 - Jan 12
Thursday, Oct 16 - Jan 22
Tues, Thurs & Sun, Dec 9 - Jan 18 (meets three times a week!)

MATHCOUNTS/AMC 8 Advanced
Tues & Thurs, Jul 8 - Aug 14 (meets twice a week!)
Sunday, Aug 17 - Nov 9
Tuesday, Aug 26 - Nov 11
Thursday, Sep 4 - Nov 20
Friday, Sep 12 - Dec 12
Monday, Sep 15 - Dec 8
Sunday, Oct 5 - Jan 11
Tues, Thurs & Sun, Dec 2 - Jan 11 (meets three times a week!)
Mon, Wed & Fri, Dec 8 - Jan 16 (meets three times a week!)

AMC 10 Problem Series
Tues & Thurs, Jul 8 - Aug 14 (meets twice a week!)
Sunday, Aug 10 - Nov 2
Thursday, Aug 14 - Oct 30
Tuesday, Aug 19 - Nov 4
Mon & Wed, Sep 15 - Oct 22 (meets twice a week!)
Mon, Wed & Fri, Oct 6 - Nov 3 (meets three times a week!)
Tue, Thurs & Sun, Oct 7 - Nov 2 (meets three times a week!)

AMC 10 Final Fives
Friday, Aug 15 - Sep 12
Sunday, Sep 7 - Sep 28
Tuesday, Sep 9 - Sep 30
Monday, Sep 22 - Oct 13
Sunday, Sep 28 - Oct 19 (1:00 - 2:30 pm ET/10:00 - 11:30 am PT)
Wednesday, Oct 8 - Oct 29
Thursday, Oct 9 - Oct 30

AMC 12 Problem Series
Wednesday, Aug 6 - Oct 22
Sunday, Aug 10 - Nov 2
Monday, Aug 18 - Nov 10
Mon & Wed, Sep 15 - Oct 22 (meets twice a week!)
Tues, Thurs & Sun, Oct 7 - Nov 2 (meets three times a week!)

AMC 12 Final Fives
Thursday, Sep 4 - Sep 25
Sunday, Sep 28 - Oct 19
Tuesday, Oct 7 - Oct 28

AIME Problem Series A
Thursday, Oct 23 - Jan 29

AIME Problem Series B
Tuesday, Sep 2 - Nov 18

F=ma Problem Series
Tuesday, Sep 16 - Dec 9
Friday, Oct 17 - Jan 30

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, Aug 14 - Oct 30
Sunday, Sep 7 - Nov 23
Tuesday, Dec 2 - Mar 3

Intermediate Programming with Python
Friday, Oct 3 - Jan 16

USACO Bronze Problem Series
Wednesday, Sep 3 - Dec 3
Thursday, Oct 30 - Feb 5
Tuesday, Dec 2 - Mar 3

Physics

Introduction to Physics
Tuesday, Sep 2 - Nov 18
Sunday, Oct 5 - Jan 11
Wednesday, Dec 10 - Mar 11

Physics 1: Mechanics
Sunday, Sep 21 - Mar 22
Sunday, Oct 26 - Apr 26
0 replies
jwelsh
Jul 1, 2025
0 replies
USAMTS and AMC 10?
HungryCalculator   1
N 10 minutes ago by neeyakkid23
Say you take the USAMTS and AMC10, and you qualify for AIME through both pathways.

Does your JMO qualification now depend on your AMC10 pathway (AMC + AIME), or just the 9+ on AIME required for JMO Qual through the USAMTS + AIME pathway?

Are you even allowed to take both USAMTS and AMC10 in the first place?
1 reply
HungryCalculator
an hour ago
neeyakkid23
10 minutes ago
Triangle in Space With Each of Its Sides Tangent to a Sphere
P_Groudon   55
N 2 hours ago by fruitmonster97
Source: 2019 AMC 10A #21 / 2019 AMC 12A #18
A sphere with center $O$ has radius $6$. A triangle with sides of length $15, 15,$ and $24$ is situated in space so that each of its sides is tangent to the sphere. What is the distance between $O$ and the plane determined by the triangle?

$
\textbf{(A) }2\sqrt{3}\qquad
\textbf{(B) }4\qquad
\textbf{(C) }3\sqrt{2}\qquad
\textbf{(D) }2\sqrt{5}\qquad
\textbf{(E) }5\qquad
$
55 replies
P_Groudon
Feb 8, 2019
fruitmonster97
2 hours ago
Show positivity and boundedness
toct0   1
N 5 hours ago by toct0
Hello I want to show positivity and boundedness of the following non-linear system, how do I proceed? All parameters should be assumed as positive. Thanks.

```
$\frac{{d}N}{{d}t} ~=~ rN\left(1 - \frac{N}{K}\right) - \frac{a(1 - m)NP}{1 + \gamma(1 - m)N}$
$\frac{{d}P}{{d}t} ~=~ \frac{b(1 - m)NP}{1 + \gamma(1 - m)N} - cP$
$\frac{{d}K}{{d}t} ~=~ \frac{(K - \alpha)(p - K)}{q + K}$
```
1 reply
toct0
5 hours ago
toct0
5 hours ago
A tool to numerically solve algebraic equations (including nonlinear systems)
vasg077   5
N Today at 6:12 AM by vasg077
Hi all,

I’m an engineering student who recently built an iOS app called Numerical Solver to help with solving algebraic equations numerically — including nonlinear equations and systems of equations. It works offline and now includes a new plotting feature that lets you visualize equations or input your own (x, y) data.

This isn’t a symbolic solver — it won’t give you step-by-step solutions — but it’s useful for checking results, exploring behavior, or tackling equations that can’t be solved analytically.

If anyone’s curious or wants to try it out, I’d be happy to share the App Store link in the comments.

I’d also love any feedback or ideas for improving it.

Thanks!
5 replies
vasg077
Yesterday at 10:17 AM
vasg077
Today at 6:12 AM
USAJMO 2025/4
LostDreams   62
N Today at 5:58 AM by lightsynth123
Source: USAJMO 2025/4
Let $n$ be a positive integer, and let $a_0,\,a_1,\dots,\,a_n$ be nonnegative integers such that $a_0\ge a_1\ge \dots\ge a_n.$ Prove that
\[
\sum_{i=0}^n i\binom{a_i}{2}\le\frac{1}{2}\binom{a_0+a_1+\dots+a_n}{2}.
\]Note: $\binom{k}{2}=\frac{k(k-1)}{2}$ for all nonnegative integers $k$.
62 replies
LostDreams
Mar 21, 2025
lightsynth123
Today at 5:58 AM
sequence
am_11235...   3
N Today at 5:24 AM by Alphaamss
Let $f:[a,b]\to [a,b]$ be a function such that $|f(x)-f(y)|\leq |x-y|$ for all $x,y\in [a,b]$. Prove that the sequence $\{x_n\}$ defined by $x_{n+1}=\frac{x_n+f(x_n)}{2}$ converges to a fixed point of $f$.
3 replies
am_11235...
Mar 19, 2017
Alphaamss
Today at 5:24 AM
Clarification about area of region bounded by curves
pie854   1
N Yesterday at 6:09 PM by greenturtle3141
Suppose we want to find the area of the region defined by $a_1\leq xy\leq a_2$ and $b_1\leq x/y \leq b_2$. Assuming $a_1,b_1>0$, I think there are going to be two such regions with equal area (corresponding to $x,y>0$ and $x,y<0$).

When I use a transition map and change the variables, the value of the integral I get is only half of the actual area (corresponding to $x,y>0$?). So, what I want to know is: is the formula $$\iint_{R} f(x,y) \ dx \ dy=\iint_{R'} f(u,v) |det(J_T)| \ du \ dv$$implicitly assuming there's only one region? And, how do I use it so that it always gives the complete region, without me having to look at the graphs?

P.S. If I let $u=xy$, $v=x/y$ then $(x,y)=(\pm \sqrt{uv}, \pm \sqrt{u/v})$. So, I think I have to consider two maps $T_1(u,v)=(\sqrt{uv},\sqrt{u/v})$ and $T_2(u,v)=(-\sqrt{uv},-\sqrt{u/v})$ and then add up the two integrals $T_1$ and $T_2$ give. Is it alright?
1 reply
pie854
Yesterday at 11:31 AM
greenturtle3141
Yesterday at 6:09 PM
a sufficient condition for nilpotence
CatalinBordea   3
N Yesterday at 4:26 PM by Filipjack
Source: Romanian District Olympiad 2015, Grade XII, Problem 4
Let $ m $ be a non-negative ineger, $ n\ge 2 $ be a natural number, $ A $ be a ring which has exactly $ n $ elements, and an element $ a $ of $ A $ such that $ 1-a^k $ is invertible, for all $ k\in\{ m+1,m+2,...,m+n-1\} . $
Prove that $ a $ is nilpotent.
3 replies
CatalinBordea
Sep 26, 2018
Filipjack
Yesterday at 4:26 PM
Need Urgent Guidance to Switch from B.Tech CSE to Math Abroad (India)
ThalapathyMath   0
Yesterday at 4:13 PM
Source: need advice
Hi everyone,
I’m Shreyan Raj, 18, from India. I’m in B.Tech CSE right now, but I’ve realized my true passion is pure mathematics. I even learned basic C and Python before reaching this decision — so this isn’t sudden. I love problem-solving, logic, derivations — not programming.

I scored 295/300 in math in 12th grade, and I want to apply for a BSc math degree abroad, preferably in affordable countries like Russia, Czech Republic, Germany, etc.

But I’m facing:

1. Family pressure not to drop or switch

2. Financial struggles (can’t afford big application fees)

3. I only have soft copies of documents

4. This July is my only chance, but I also have internal exams

Can anyone here suggest:

1. Universities that accept soft docs?

2. Anyone who did something similar?

3. How to explain to parents or find support?

Even small advice means a lot right now. Thank you in advance
— Shreyan Raj

0 replies
ThalapathyMath
Yesterday at 4:13 PM
0 replies
A^2+B^2=AB+BA
mathisreal   1
N Yesterday at 10:50 AM by loup blanc
Source: OIMU 2023 #4
Determine all pairs of real matrices $(A,B)$ of size $2\times 2$ with $A\neq B$ such that
\[A^2+B^2=AB+BA=2I\]
1 reply
mathisreal
Yesterday at 12:03 AM
loup blanc
Yesterday at 10:50 AM
ISI UGB 2025 P3
SomeonecoolLovesMaths   14
N Yesterday at 10:10 AM by Cats_on_a_computer
Source: ISI UGB 2025 P3
Suppose $f : [0,1] \longrightarrow \mathbb{R}$ is differentiable with $f(0) = 0$. If $|f'(x) | \leq f(x)$ for all $x \in [0,1]$, then show that $f(x) = 0$ for all $x$.
14 replies
SomeonecoolLovesMaths
May 11, 2025
Cats_on_a_computer
Yesterday at 10:10 AM
ISI UGB 2025 P1
SomeonecoolLovesMaths   10
N Yesterday at 8:03 AM by Cats_on_a_computer
Source: ISI UGB 2025 P1
Suppose $f \colon \mathbb{R} \longrightarrow \mathbb{R}$ is differentiable and $| f'(x)| < \frac{1}{2}$ for all $x \in \mathbb{R}$. Show that for some $x_0 \in \mathbb{R}$, $f \left( x_0 \right) = x_0$.
10 replies
SomeonecoolLovesMaths
May 11, 2025
Cats_on_a_computer
Yesterday at 8:03 AM
2024 Putnam A1
KevinYang2.71   23
N Yesterday at 12:05 AM by megahertz13
Determine all positive integers $n$ for which there exists positive integers $a$, $b$, and $c$ satisfying
\[
2a^n+3b^n=4c^n.
\]
23 replies
KevinYang2.71
Dec 10, 2024
megahertz13
Yesterday at 12:05 AM
Matrices satisfy three conditions
ThE-dArK-lOrD   4
N Monday at 7:53 PM by Adustat
Source: IMC 2019 Day 1 P5
Determine whether there exist an odd positive integer $n$ and $n\times n$ matrices $A$ and $B$ with integer entries, that satisfy the following conditions:
[list=1]
[*]$\det (B)=1$;[/*]
[*]$AB=BA$;[/*]
[*]$A^4+4A^2B^2+16B^4=2019I$.[/*]
[/list]
(Here $I$ denotes the $n\times n$ identity matrix.)

Proposed by Orif Ibrogimov, ETH Zurich and National University of Uzbekistan
4 replies
ThE-dArK-lOrD
Jul 31, 2019
Adustat
Monday at 7:53 PM
Invert Your Expectations
AwesomeYRY   44
N Jun 2, 2025 by akliu
Source: USAMO 2022/3
Let $\mathbb{R}_{>0}$ be the set of all positive real numbers. Find all functions $f:\mathbb{R}_{>0} \to \mathbb{R}_{>0}$ such that for all $x,y\in \mathbb{R}_{>0}$ we have
\[f(x) = f(f(f(x)) + y) + f(xf(y)) f(x+y).\]
44 replies
AwesomeYRY
Mar 24, 2022
akliu
Jun 2, 2025
Invert Your Expectations
G H J
G H BBookmark kLocked kLocked NReply
Source: USAMO 2022/3
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
AwesomeYRY
579 posts
#1 • 10 Y
Y by v4913, MathLuis, fluff_E, HWenslawski, megarnie, son7, GoodMorning, michaelwenquan, jk2845, aidan0626
Let $\mathbb{R}_{>0}$ be the set of all positive real numbers. Find all functions $f:\mathbb{R}_{>0} \to \mathbb{R}_{>0}$ such that for all $x,y\in \mathbb{R}_{>0}$ we have
\[f(x) = f(f(f(x)) + y) + f(xf(y)) f(x+y).\]
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
pad
1671 posts
#2 • 24 Y
Y by fluff_E, HWenslawski, Aryan-23, son7, Kobayashi, AlphaBetaGammaOmega, michaelwenquan, mijail, PHSH, rama1728, skyguy88, Hoto_Mukai, Iora, crazyeyemoody907, rg_ryse, thanosaops, math12345678, hakN, rayfish, gpen1000, aidan0626, Sedro, MathRook7817, qwedsazxc
We have $f(x)>f(f^2(x)+y)$. We'll derive a LOT of information from this alone. Equivalently, this says:
\[ C>f^2(x) \implies f(C)<f(x) \quad \forall C,x. \qquad (\spadesuit)\]If $x>f^2(x)$, then $C=x$ in $(\spadesuit)$ gives $f(x)<f(x)$, contradiction. Hence $\boxed{f^2(x) \ge x}$. So $f^2(x) \le f^4(x) \le f^6(x) \le \cdots$ by iterating. Also $f(x) \le f^3(x) \le f^5(x) \le \cdots$.

The contrapositive of $(\spadesuit)$ is very useful:
\[ f(C) \ge f(x) \implies C \le f^2(x) \quad \forall C,x. \qquad (\heartsuit)\]Since $f^5(x)\ge f(x)$, plugging in $C=f^4(x)$ into $(\heartsuit)$ gives $f^4(x) \le f^2(x)$. But recall $f^4(x) \ge f^2(x)$ too, so $\boxed{f^2(x)=f^4(x)}$.

Claim: $f$ is injective.
Proof: Suppose $f(a)=f(b)$ and $b>a$. Now we utilize the entire FE. Plug in $x=f^2(n)$ into the original FE for some arbitrary real $n$:
\begin{align*}
f^3(n) &= f(f^4(n)+y) + f(f^2(n)\cdot f(y))\cdot f(f^2(n)+y) \\
&= f(f^2(n)+y) \cdot [1+f(f^2(n)\cdot f(y))]. 
\end{align*}Plug in $y=a$ and $y=b$ into the above and compare. Then we get $f(f^2(n) + a) = f(f^2(n) + b)$.

Now the original FE with $x=n$ reads:
\[ f(n) = f(f^2(n) + y ) + f(nf(y))f(n+y). \]Plugging in $y=a$ and $y=b$ and comparing gives $f(n+a)=f(n+b)$.

This holds for arbitrary $n$, so $f(x+a)=f(x+b)$ for all $x$. Now iterating and chaining this, we get $f(x+a) = f(x+B)$ for arbitrarily large $B$ since $b>a$. Take large enough $B$ for which $x+B>f^2(x+a)$. Then by $(\spadesuit)$, $f(x+B) < f(x+a)$, contradiction. Hence $b=a$. $\blacksquare$

Remember $f^2(x)=f^4(x)$. So now by injectivity, $\boxed{f^2(x)=x}$. Now $(\heartsuit)$ reads to give $f$ decreasing. The FE is now: $f(x)=f(x+y)[1+f(xf(y))]$. Plugging in $x=1$ gives $f(y+1)=f(1)/(y+1)$. So $f(x)=C/x$ for $x>1$ for some constant $C$. So since $f$ is an involution, $f(x)=C/x$ for $x<C$ too. Now we want $f$ for sub-1 inputs. Fix $x\le 1$, and plug in $y=100$ into the new FE, and we get $f(x)=C/x$. Now we can confirm $f(x)=C/x$ works for all $x>0$, and it must be the solution form.

Remarks
This post has been edited 3 times. Last edited by pad, Mar 24, 2022, 10:50 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
CANBANKAN
1302 posts
#3 • 8 Y
Y by Smartgoodness, fluff_E, soggydollar, son7, michaelwenquan, Shiro2911, rama1728, EthanWYX2009
The answer is $f(x)\equiv \frac ax$ which clearly works. Let $P(x,y)$ denote the assertion in the question.

Claim 1 $x\le f(f(x))$.

Proof: Assume not, then $P(x,x-f(f(x)))$ gives $f(x)=f(x)$+smth nonzero, contradiction.

Now, notice $f(x)\le f(f(f(x)))$ but $f(x)>f(f(f(x))+y)$ for any $y\in \mathbb{R}^+$

Claim 2: If $f(f(x))>f(f(y))$ then $f(x)<f(y)$.

Proof: $P(y,f(f(x))-f(f(y)))$ gives $f(y)>f(f(f(x)))\ge f(x)$.

Claim 3: $f(f(x))=f(f(f(f(x))))$ for all $x\in \mathbb{R}^+$

Proof: Assume not. Then $f(f(x))<f(f(f(f(x))))$ and $f(x)<f(f(f(x)))$. Let $A=f(f(x))$, then we have $f(x)<f(A)$ and $f(f(x))<f(f(A))$, but $f(f(x))<f(f(A))$ implies $f(x)>f(A)$, absurd.

Claim 4: $f(x)<\frac{f(1)}{x-1}$

Proof: $P(1,x-1)$.

Claim 5:$f(x)\equiv f(f(f(x)))$ for all $x\ge 1$

Let $B=f(f(x))$. We are given $f(f(x))=f(f(B))$ by claim 3.

Assume $f(x)<f(B)$ since $f(x)\le f(f(f(x)))$. This also means $1\le x<B.$ Then $P(x,y), P(B,y)$ yields $$f(x)=f(f(f(x))+y)+f(x+y)f(xf(y))$$
$$f(B)=f(f(f(B))+y)+f(B+y)f(Bf(y))$$
Let $C=f(B)-f(x)$. Plugging in $y=f(t)$ gives $f(B)-f(x)< f(B+f(t))f(Bf(f(t))) <\frac{f(1)}{B-1} \cdot \frac{f(1)}{Bf(f(t))-1} \le \frac{f(1)}{B-1} \cdot \frac{f(1)}{Bt-1}$ which tends to 0 as $t$ tends to infinity.

Claim 6 if $x\ge 1$ then $f(f(x))> f(f(x)+y)$ for any $y\in \mathbb{R}^+$.

Proof: $P(f(x),y)$ gives $f(f(x))=f(f(x)+y)+f(f(x)+y)f(f(x)f(y))$.

Also, from $P(f(f(x)),y)-P(x,y)$ we get for all $x\in \mathbb{R}_{\ge 1}, y\in \mathbb{R}^+$, $$f(f(f(x))+y)f(f(f(x))f(y)) = f(x+y)f(xf(y))$$is an identity, call $Q(x,y)$.

Claim 7 $f(f(1))=1$.

If $x=1,$ and $f(f(1))>1$ we have

$f(f(y))>f(f(f(1)) \cdot f(y))$ for all $y\in \mathbb{R}_{\ge 1}$ because $f(f(1))\cdot f(y)$ and claim 6.

Therefore, we select $y$ such that $1+y=f(f(z))$ where $z\ge 2$. We can see that $f(1+y)=f(f(f(z)))>f(k)$ for any $k>f(f(z))$. Therefore, $f(1+y)>f(f(f(1))+y)$.

Multiplying, $f(f(y))f(1+y) > f(f(f(1))+y) f(f(f(1))f(y))$ for this $y\in \mathbb{R}_{\ge 1}$

Claim 8 $f(1+y)=f(1+f(f(y)))$ for all $y\ge 1$

$P(1,y)$ yields $f(1)=f(1+y)(1+f(f(y)))$

$P(1,f(y))$ yields $f(1)=f(1+f(y))(1+f(y))$ where $y\ge 1$. Thus, for $y\ge 1, f(1+f(y))=\frac{f(1)}{1+f(y)}$

In particular, $f(1+f(f(y)))=\frac{f(1)}{1+f(f(y))}=f(1+y)$

Claim 9 Let $a=f(1)$, then for all $x\ge 1$, $f(x+a)=f(f(f(x))+a)$.

Proof: $Q(x,a)$ gives $f(f(f(x))+a)f(f(f(x)))=f(x+a)f(x)$ since $f(a)=1$. Since $f(f(f(x)))=f(x)$, the conclusion follows.

Now, $x$ denotes a real that is at least 1.

Claim 10 $f(x)f(af(x))=1$ for all $x\ge 1$

$P(x,a)$ gives $f(x)=f( f(f(x))+a)+f(x+a)f(x)=f(x+a)(1+f(x))$.

Therefore, $f(x+a)=\frac{f(x)}{1+f(x)}$

$P(a,x)$ gives $f(a)=f(f(f(a))+x)+f(a+x)f(af(x))=f(a+x)(1+f(af(x))$

Therefore, $1=\frac{f(x)}{1+f(x)}(1+f(af(x)))$

$1+f(x)=f(x)(1+f(af(x))$

$1=f(x)f(af(x))$ for all $x\ge 1$

Claim 11 For all $x\ge 1$, $f(x)f(f(x))=a$.

$P(1,x)$ gives $a=f(1+x)+f(1+x)f(f(x))$. Therefore, for all $x\ge 0$, $f(1+x)=\frac{a}{1+f(f(x))}$

In particular, $f(1+f(x))=\frac{a}{1+f(x)}$ if $x\ge 1$.

Now, $P(f(x),1)$ where $x\ge 1$ gives

$f(f(x))=f(f(x)+1)+f(f(x)+1)f(af(x))$

$f(f(x))=\frac{a}{1+f(x)} (1+f(af(x)))$.

By Claim 10, if $x\ge 1$ then $f(af(x))=\frac{1}{f(x)}$

Therefore, $f(f(x))=\frac{a}{1+f(x)} (1+\frac{1}{f(x)}) = \frac{a}{1+f(x)} \frac{1+f(x)}{f(x)}$, so $f(f(x))f(x)=a$. In particular, $f(x)\le \frac ax$

By Claim 11, for all $x\ge 1$, we have $f(x)=\frac{a}{f(f(x))}\le \frac ax$ and if $f(x)=\frac ax$ then it follows that $f(f(x))=x$.

This means $f(x)=\frac ax$ for a dense set in $\mathbb{R}_{\ge 1}$, call $S$. This set $S$ satisfies for all $t\in S, f(f(t))=t$. We note this set $S$ is also $Im(f)=Im(f(f(x))$. Indeed, if $f(t)=u$ then $uf(u)=a$ and if $uf(u)=a$ then $u=f(\frac au)$

Claim 12 $f(xf(y))=f(f(f(x))f(y))$

By $Q(x,f(y))$, we have $f(f(f(x))+f(y))f(f(f(x))f(f(y)))=f(x+f(y))f(xf(f(y)))$.

By $P(f(x),y)-P(f(x),f(f(y))$, we can get $f(f(x)+y)=f(f(x)+f(f(y))) = \frac{f(f(x))}{1+f(f(x)f(y))}$. Therefore, $f(f(f(x))+f(y))=f(x+f(y))$, it follows that $f(xf(f(y)))=f(f(f(x))f(f(y)))$. Since the range of $f(f(y))$ is precisely the range of $f(y)$, it follows that $f(xf(y))=f(f(f(x))f(y))$

Therefore, $f(xf(y))=\frac{f(x)}{f(y)}=\frac{f(f(f(x)))}{f(y)}=f(f(f(x))f(y))$.

Now, $Q(x,y)$ gives $f(x+y)=f(f(f(x))+y)$ for all $x\ge 1, y\in \mathbb{R}^+$. Picking $y=f(f(z))-x$ gives a contradicting $f(t)>f(u)$ if $u>t$ and $t\in S$. Therefore, it follows that $f(f(x))=x$ for all $x\ge 1$. Since $S$ is closed under inverses, $S=\mathbb{R}^+$, as desired.
This post has been edited 4 times. Last edited by CANBANKAN, Apr 5, 2022, 1:16 AM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
TLP.39
778 posts
#4 • 1 Y
Y by fluff_E
Let $P(x,y)$ denotes the equation.

What I got during the test

Continuation

@below Yep. On another note, I didn't see that the method in claim 2 works. That's quite a good job, I suppose
This post has been edited 3 times. Last edited by TLP.39, Mar 24, 2022, 6:34 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
CANBANKAN
1302 posts
#5 • 1 Y
Y by fluff_E
wow I made more progress than you during the test. I proved $f(x)=f^3(x)$ :O

@above I didn't realize injectivity works :wallbash:
that would've cut down my solution length to half
This post has been edited 1 time. Last edited by CANBANKAN, Mar 24, 2022, 6:56 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
USJL
540 posts
#6 • 21 Y
Y by ike.chen, TLP.39, GorgonMathDota, 62861, eagles2018, lethan3, Inconsistent, dchenmathcounts, MathLuis, L567, Rg230403, SnowPanda, son7, asimov, Wizard0001, rayfish, Aritra12, PHSH, rg_ryse, Aryan-23, OronSH
My problem! :D

The main trickery is basically if we have the condition $f(x)>f(f(f(x))+y)$, then surprisingly we can prove $f^{(4)}(x)=f^{(2)}(x)$!!!

The two solutions above more or less do the same thing, but I didn't notice that there's some way around to get $f^{(3)}(x)=f(x)$ (as you can see, I have the trick in mind when composing this problem so it became extremely hard for me to see other solutions lol).

I think the solution in #4 is really nice, but for those who only got $f^{(4)}(x)=f^{(2)}(x)$, here is one way to get injectivity:

If $a>b$ but $f(a)=f(b)$, consider $P(f^{(2)}(x), a)$ and $P(f^{(2)}(x), b)$. By comparing them, we get $f(f^{(2)}(x)+a)=f(f^{(2)}(x)+b)$.

Now plugging this in back to $P(x,a), P(x,b)$, and we get $f(x+a)=f(x+b)$. But periodicity is clearly absurd---for example, take $x=f^{(2)}(a)-b$ (which is positive as $f^{(2)}(a)\geq a$. Therefore $f$ is injective.

(Credit to FEcreator for this clean injectivity argument.)

From here, one can really do anything to finish the solution.

Some proposer remark
This post has been edited 1 time. Last edited by USJL, Mar 24, 2022, 10:26 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
GeronimoStilton
1521 posts
#7 • 1 Y
Y by ayana350
We claim $f(x) = \frac cx$ for some constant $c>0$. It is clear that $f$ works. Now consider some working function $f$.

Let $P(x,y)$ denote the condition. Abbreviate $f(f(f(x)))$ to $f^3(x)$ etc. for convenience. Observe that for $t>f(f(x))$, taking $P(x,t-f(f(x)))$ yields $f(x)>f(t)$. In particular, we must have $f(f(x)) \ge x$ for all $x$ because we otherwise get $f(x)> f(x)$, absurd.

Claim: For all $x$, we have $f^4(x) = f^2(x)$.

Solution: Suppose otherwise. Then for some $x$, we have $f^4(x) > f^2(x)$. Then we get $f^5(x) < f(x)$, which is absurd because $f^3(x)\ge f(x)$ and $f^5(x)\ge f^3(x)$, so we are done. $\fbox{}$

Taking $P(f^2(x),y)$ yields
\[f^3(x) = f(f^2(x)+y)+f(f^2(x)f(y))f(f^2(x)+y) = f(f^2(x)+y)\cdot [1+f(f^2(x)f(y))]\qquad (\diamondsuit).\]Observe that $P(1,y)$ yields
\[f(1) = f(f(f(1))+y)+f(f(y))f(1+y)\ge f(f(f(1))+y)+yf(1+y).\]Hence we have the bound
\[\frac{f(1)}{y} > f(1+y) \qquad (\clubsuit).\]
Claim: $f$ is injective.

Solution: Suppose otherwise. Then there exist some $a<b$ with $f(a)=f(b)$. By $\diamondsuit$, this implies that for all $x$ we have $f(f^2(x)+a)=f(f^2(x)+b)$. In turn, taking $P(x,a)$ and $P(x,b)$ yields $f(xf(a))f(x+a) = f(xf(b))f(x+b)$, so $f(x+a)=f(x+b)$. Let $k$ be the minimum positive integer such that $f(1)10^{-k} < f(a+1)$. Define $N = \lceil \frac{10^k}{b-a}\rceil$ to yield
\[f(1)10^{-k}<f(a+1) = f(a+1+(b-a)) = \cdots = f(a+1+N(b-a)) < \frac{f(1)}{N(b-a)} \le f(1)10^{-k}\]by $\clubsuit$, absurd. $\fbox{}$

Then $f^4(x) = f^2(x)$ implies $f^3(x) = f(x)$ and $f^2(x)=x$, so $f$ is surjective and the given equation rewrites as
\[f(x)=f(x+y)+f(xf(y))f(x+y) = f(x+y)\cdot [1+f(xf(y))].\]Take $P(1,y)$ to yield $f(1) = f(1+y)\cdot [1+f(f(y))] = (1+y)f(1+y)$. Thus for $x > 1$, we have $f(x) = \frac 1x f(1)$. Then for $x>1$, $P(x,y)$ rewrites as
\[\frac 1x f(1) = \frac{1}{x+y}f(1)\cdot [1+f(xf(y))],\]which rearranges to $\frac yx = f(xf(y))$. For $y>1$, this is $\frac yx = f(f(1)\cdot \frac xy)$. As all reals may be expressed as $\frac xy$ for some $x,y>1$, this means that $f(x)=\frac 1x f(1)$ for all $x$. This is equivalent to the stated solution set.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
megarnie
5658 posts
#8 • 3 Y
Y by GoodMorning, Spectator, OronSH
Let $\mathbb{R}_{>0}$ be the set of all positive real numbers. Find all functions $f:\mathbb{R}_{>0} \rightarrow \mathbb{R}_{>0}$ such that for all $x, y \in \mathbb{R}_{>0}$ we have
$$f(x) = f (f(f(x)) + y) + f(xf(y))f(x + y).$$.

Solved with hint from rama1728 (the solution set)

Let $P(x,y)$ denote the given assertion.

We claim the answer is $\boxed{f(x)=\frac{c}{x}}$ for some real constant $c$. It's easy to check that this works. We now prove it's the only solution.

Claim 1: If $f$ is involution, we are done.
Fixed proof thanks to jasperE3: Suppose $f(f(x))=x$ for all $x$.

$P(1,x): f(1)=f(1+x)+xf(x+1)\implies f(x+1)=\frac{f(1)}{x+1}$.

So $f(x)=\frac{f(1)}{x}$ for $x>1$. Let $f(1)=c$.

Set $x,y>1$. Note that $x>1$ and $x+y>1$. Then \[f(x)=f(x+y)+f\left(\frac{cx}{y}\right)f(x+y)=f(x+y)\left(f\left(\frac{cx}{y}\right)+1\right).\]
So $\frac{c}{x}=\frac{c}{x+y}\left(f\left(\frac{cx}{y}\right)+1\right)$.

This implies $\frac{x+y}{x}=f\left(\frac{cx}{y}\right)+1$, so $f\left(\frac{cx}{y}\right)=\frac{y}{x}$, so $f(x)=\frac{c}{x}$ for all $x>0$. $\blacksquare$

Claim 2: $f(f(x))\ge x$ for all $x$.
Proof: If, not for some $x$, then if we set $y=x-f(f(x))$, then $f(xf(y))f(x+y)=0$. $\blacksquare$

Claim 3: If $f(f(y))>f(f(x))$, then $f(x)>f(y)$.
Proof:\[P(x,f(f(y))-f(f(x)): f(x)=f^3(y)+f(xf(f(f(y))-f(f(x))))f(x+f(f(y))-f(f(x)))>f(y)\]$\blacksquare$

Claim 4: If $f(x)>f(y)$, then $f(f(y))>f(f(x))$,
Proof: $P(f(y),f(x)-f(y)): f(f(y))=f(f^3(y)+f(x)-f(y))+f(f(y)f(f(x)-f(y)))f(f(x))>f(f(x))$. $\blacksquare$

Claim 5: $f^4(x)= f^2(x)$.
Proof: Suppose for some $k$, $f^4(k)>f^2(k)$.

$P(k,f^4(k)-f^2(k)): f(k)=f^5(k)+smth>f^5(k)$, but also $f^5(k)\ge f^3(k)\ge k$, a contradiction. $\blacksquare$

Claim 6: $f^3(x)=f(x)$.
Proof: If $f^3(x)>f(x)$, then $f^4(x)<f^2(x)$, not true. $\blacksquare$


Claim 7: If $f$ is injective, we are done.
Proof: If $f$ is injective, since $f^3(x)=f(x)$, $f^2(x)=x$. $\blacksquare$

Claim 8: $f$ is injective.
Proof: $P(f(f(x)),y): f(x)=(f(f(f(x))f(y))+1)f(f(f(x))+y)$

So comparing with the original FE gives $f(xf(y))f(x+y)=f(f(f(x))f(y))f(f(f(x))+y)$ (1)

Looking at $P(f(f(x)),a)$ and $P(f(f(x)),b)$, we find that if $f(a)=f(b)$, then $f(f(f(x))+a)=f(f(f(x))+b)$.

Now looking setting $y=a$ and $y=b$ in $(1)$, we find $f(x+a)=f(x+b)$.

So for $x>a$, $f(x)=f(x+b-a)$.

If $b\ne a$, $P(x,b-a): f(x)=f(x)+smth$, where $smth>0$, a contradiction.
This post has been edited 2 times. Last edited by megarnie, Mar 25, 2022, 7:03 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
MathLuis
1594 posts
#9
Y by
Very tricky one, i love it :D.
We claim that $f(x)=\frac{c}{x}$ where $c \in \mathbb R_{>0}$ works.
Let $P(x,y)$ the assertion of the given F.E.
Claim 1: $f(f(x)) \ge x$
Proof: Assume that there exists $x$ such that $f(f(x))<x$ then by $P(x,x-f(f(x)))$
$$0=f(xf(x-f(f(x))))f(2x-f(f(x))) \; \text{contradiction!!}$$Claim 2: $f^{2k}(x)=f(f(x)) \; \forall k \in \mathbb Z_{>0}$
Proof: Assume that there exists $k \in \mathbb Z_{>0}$ and $x \in \mathbb R_{>0}$ such that $f^{2k}(x)>f(f(x))$ then by $P(x,f^{2k}(x)-f(f(x)))$ and Claim 1
$$f(x)=f^{2k+1}(x)+f(xf(f^{2k}(x)-f(f(x))))f(x+f^{2k}(x)-f(f(x))) \ge f(x)+f(xf(f^{2k}(x)-f(f(x))))f(x+f^{2k}(x)-f(f(x))) \implies 0 \ge f(xf(f^{2k}(x)-f(f(x))))f(x+f^{2k}(x)-f(f(x))) \; \text{contradiction!!}$$So $f^{2k}(x) \le f(f(x))$ but using Claim 1 we get $f^{2k}(x) \ge f(f(x))$ so $f^{2k}(x)=f(f(x))$ for any $k \in \mathbb Z_{>0}$ as desired.
Claim 3: $f$ is injective.
Proof: Assume that there exists $f(a)=f(b)$ with $a>b$ then comparing $P(f(f(x)),a)$ with $P(f(f(x)),b)$ along with Claim 2
$$f(f(f(x))+a)=f(f(f(x))+b)$$Now conparing $P(x,a)$ with $P(x,b)$ we get
$$f(x+a)=f(x+b) \implies f(x)=f(x+a-b) \; \forall x \in \mathbb R_{>b}$$Now by $P(f(x),a-b)$ and Claim 2
$$1=1+f(f(x)f(a-b)) \implies f(f(x)f(a-b))=0 \; \text{contradiction!!}$$Hence $f$ is injective as desired.
Finishing: By injectivity on Claim 2 we get $f(f(x))=x$ so $f$ is an involution and now we re-write the F.E. as
$$f(x)=(1+f(xf(y)))f(x+y)$$Call $Q(x,y)$ the assertion of this F.E., then using $Q(1,x-1)$ where $x>1$
$$f(1)=xf(x) \implies f(x)=\frac{f(1)}{x} \; \forall x \in \mathbb R_{>1}$$Now set $x>1$ and use $Q \left(x, f \left(\frac{1}{x} \right) \right)$
$$1+\frac{f\left(\frac{1}{x} \right)}{x}=1+f(1) \implies f \left(\frac{1}{x} \right)=xf(1) \; \forall x \in \mathbb R_{>1}$$Now make the replace $t=\frac{1}{x}$ so here clearly $t$ can take all the values on $[0,1]$ so we got $f(t)=\frac{f(1)}{t} \; \forall t \in [0,1]$
So now let $f(1)=c$ and along with the other result we got $$\boxed{f(x)=\frac{c}{x} \; \forall x \in \mathbb R_{>0}}$$so we are done :D
This post has been edited 2 times. Last edited by MathLuis, Dec 23, 2022, 4:32 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Serena_Xu
101 posts
#10 • 3 Y
Y by Tyler0911, PHSH, GmathIRL
I still can't solve it rip.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
MathLuis
1594 posts
#11
Y by
Serena_Xu wrote:
I still can't solve it rip.

I'm pretty sure u will soon
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
L567
1189 posts
#12
Y by
Solved with p_square

We claim the only solutions are $f(x) = \frac{c}{x}$ for some $c \in \mathbb{R}^+$, which can be checked to work. Let $P(x,y)$ denote the given assertion.

Claim 1: $f^4(x) = f^2(x)$

Proof: Taking $P(x,z-f(f(x)))$ for some $z > f(f(x))$ gives that $z > f(f(x)) \implies f(z) < f(x)$. In particular, taking $z = x$ gives that $f(f(x)) \ge x$ for all $x$.

So we know $f^4(x) \ge f^2(x)$. But if we had $f^4(x) > f^2(x)$, then taking $z = f^4(x)$, we have that $f^2(x) \ge f^4(x)$, a contradiction. So we have $f^4(x) = f^2(x)$ for all $x$. $\square$

Let $S$ denote the set of all reals $s$ such that $f(f(s)) = s$, note that $f^2(x) \in S$ by Claim 1. $P(s,y)$ gives us $f(s) = f(s+y)(f(sf(y))+1)$ for $s \in S$. Call this $Q(s,y)$.

Claim 2: $f(x) = \frac{c}{x}$ for all $x \in S$.

Proof: Let $m,n$ be two elements of $S$, note that $f(m), f(n)$ are also in $S$. Dividing $Q(m,f(n))$ by $Q(n,f(m))$ and comparing it with dividing $Q(f(m), n)$ by $Q(f(n), m)$, we obtain $$\frac{f(m)}{f(n)} = \frac{f(m+f(n))}{f(n+f(m))} = \frac{n}{m}$$so the value $mf(m)$ is constant over $S$ and so there exists some $c$ such that $f(x) = \frac{c}{x}$ for all $x \in S$. $\square$

Claim 3: $f$ is injective.

Proof: Suppose not, and $f(a) = f(b)$ with $a \neq b$. Then taking $y = a$ and $y = b$ in the previous equation, we get $f(s+a) = f(s+b)$.

Comparing $P(x,a)$ and $P(x,b)$, we get $$f(f(f(x)) + a) + f(x+a)f(xf(a)) = f(f(f(x))+b) + f(x+b)f(xf(b))$$but by the previous paragraph, since $f(f(x)) \in S$, we must have that $f(f(f(x)) + a) = f(f(f(x))+b)$ so $f(x+a) = f(x+b)$ for all positive reals $x$. So this means $f$ is periodic with period $p = |a-b| > 0$ for all $x > \min(a,b)$. But then $P(x,x+np-f(f(x)))$ for sufficiently large $n$ and $x > a+b$ gives a contradiction, so $f$ is indeed injective. $\square$

Since $f(f(x)) = f(f(f(f(x))))$ and $f$ is injective, this means that $f(f(x)) = x$ for all $x$ so $S  = \mathbb{R}^+$, since we had $f(x) = \frac{c}{x}$ for all $x \in S$, it follows that $f(x) = \frac{c}{x}$ for all $x \in \mathbb{R}^+$, so we are done. $\blacksquare$
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
dchenmathcounts
2443 posts
#13
Y by
AAAAAGH. This is about the same difficulty as That One USEMO FE, a little disappointed I didn't get this one in contest. edit: lolno I was just being bad

The answer is all functions of the form $f(x)=\frac{c}{x}$ for constant $c$. Check they work.

Note $f^2(x)\geq x$ as otherwise $y=x-f^2(x)$ gives $f(x)=f(x)+f(xf(y))f(x+y)>f(x)$, contradiction.

If $f^2(a)>f^2(b)$ we claim $f(a)<f(b)$. Plugging in $x=b$ and $y=f^2(a)-f^2(b)$ gives
\[f(b)=f^3(a)+f(xf(y))f(x+y)>f^3(a)\geq f(a)\]as desired.

Then we claim $f^2(x)=f^4(x)$ for all $x$. If not then $f^2(x)<f^4(x)$, implying $f(x)>f^3(x)$ due to above, contradiction as $f(x)\leq f^3(x)$.

Now we prove $f$ is injective. Let $f(a)=f(b).$ Then
\[f^3(x)=f(f^4(x)+a)+f(f^2(x)f(a))f(f^2(x)+a)\]implies
\[\frac{f^3(x)}{1+f(f^2(x)f(a))}=f(f^2(x)+a)\]Note LHS is just in terms of $x$ and $f(a)$, so $f(f^2(x)+a)=f(f^2(x)+b)$.

Now $x,a$ gives
\[f(x)=f(f^2(x)+a)+f(xf(a))f(x+a)\]so $f(x+a)=f(x+b)$.

Let $b\geq a$ and $b-a=d$; then for all $x>a,$ we have $f(x)=f(x+kd)$ for integer $k$. This means that $f^2(x+kd)=f^2(x)\geq x+kd$ for any integer value of $k$. This is only possible if $d=0$; otherwise, $x+kd$ is unbounded while $f^2(x)$ is a constant. Thus $a=b$.

Injectivity then gives $x=f^2(x)$. So substituting $x=1$ gives $\frac{f(1)}{y+1}=f(y+1),$ or $f(x)=\frac{f(1)}{x}$ for $x>1.$

Now we prove this for numbers less than $1$. Fix $x>1$ and then take some $y>1$ such that $xf(y)<1$, taking advantage of the fact that $f$ can be arbitrarily small. Then
\[f(x)=f(x+y)(1+f(xf(y))\]\[\frac{f(1)}{x}=\frac{f(1)}{x+y}(1+f(xf(y))\]\[\frac{y}{x}=f(xf(y))=f(\frac{xf(1)}{y})\]
Obviously $\frac{xf(1)}{y}$ can take on any real between $0$ and $1$. If we let $\frac{xf(1)}{y}=r$ then the equation can be rewritten as $\frac{f(1)}{r}=f(r)$ as desired.
This post has been edited 2 times. Last edited by dchenmathcounts, Mar 25, 2022, 3:01 AM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
USJL
540 posts
#14
Y by
dchenmathcounts wrote:
AAAAAGH. This is about the same difficulty as That One USEMO FE, a little disappointed I didn't get this one in contest.

The answer is all functions of the form $f(x)=\frac{c}{x}$ for constant $c$. Check they work.

Note $f^2(x)\geq x$ as otherwise $y=x-f^2(x)$ gives $f(x)=f(x)+f(xf(y))f(x+y)>f(x)$, contradiction.

Now we prove $f$ is injective. Let $f(a)=f(b).$ Then
\[f(a)=f(f^2(a)+y)+f(af(y))f(a+y)\]implies
\[f(a+y)=\frac{f(a)-f(f^2(a)+y)}{f(af(y))}.\]Note RHS is just in terms of $a$, so $f(a+y)=f(b+y)$.

Hmph, why is $f(a+y)=f(b+y)$? How did you deal with $f(af(y)), f(bf(y))$?

Also I think your argument for $f^{(3)}(x)=f(x)$ is wrong. You would only get $f^{(4)}(x)=f^{(2)}(x)$.
This post has been edited 2 times. Last edited by USJL, Mar 25, 2022, 2:36 AM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
dchenmathcounts
2443 posts
#15
Y by
Argh, yaeh that is wrong. I mistakenly thought all of RHS was just wrapped around $f(a)$ when the denominator isn't.

So injectivity probably looks like everyone else's solution, the $f^2(x),y$ substitution and letting $y$ be $a,b$ (since my solution gets $f=f^3$ independently of injectivity edit: apparently not, then f^3(x),y?). And then you get $f(f^2(x)+y)$ is a constant as long as $f(y)$ is

(Will make edits to correct my solution)

edit to edits: Yeah, I needed f^2(x),y before the f(x+a)=f(x+b) argument
This post has been edited 3 times. Last edited by dchenmathcounts, Mar 25, 2022, 3:02 AM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
oVlad
1746 posts
#17
Y by
The solution is pretty smooth! Let $P(x,y)$ denote the given assertion, and let $f^k(x)$ denote the $k^{\text{th}}$ composition of $f.$ Since the codomain of $f$ is $\mathbb{R}_{>0},$ from $P(x,y)$ it follows that $f(x)>f(f^2(x)+y)$ for all $x,y\in\mathbb{R}_{>0}.$ Let this property be $Q(x,y).$

Claim 1: For all $x\in\mathbb{R}_{>0}$ we have $f^4(x)=f^2(x).$

Proof: Firstly notice that if $f^2(x)<x$ for some $x,$ then $Q(x,f^2(x)-x)$ yields $f(x)>f(x),$ which is absurd. Thus, we can conclude that $f^2(x)\geq x$ for all $x\in\mathbb{R}_{>0}.$ In particular, this implies that $f^4(x)\geq f^2(x).$

Assuming that $f^4(x)\neq f^2(x),$ then $Q(x,f^4(x)-f^2(x))$ gives us $f^5(x)<f(x).$ This is yet again absurd, since we have $f(x)\leq f^3(x)\leq f^5(x)$ as per our previous observation. Thus, for all $x\in\mathbb{R}_{>0}$ we have $f^4(x)=f^2(x).$ $\blacksquare$

Claim 2: The function $f$ is injective.

Proof: Assume, for the sake of contradiction, that $f(a)=f(b)$ for some $a\neq b.$ We say that a number $x$ is good if $x=f^2(y)$ for some $y.$ Claim 1 implies that $f^2(x)=x$ for all good numbers. By comparing $P(y,a)$ and $P(y,b)$ for some good $y$ we can deduce that $f(y+a)=f(y+b).$

However, by looking at $P(y,a)$ and $P(y,b)$ for any $y$ (not necessarily a good $y$ this time) we get that \begin{align*}f(y)&=f(f^2(y)+a)+f(yf(a))\cdot f(y+a) \\ &=f(f^2(y)+b)+f(yf(b))\cdot f(y+b)\end{align*}Notice that $f^2(y)$ is good, and thus $f(f^2(y)+a)=f(f^2(y)+b)$ (as per our argument on the previous line). Therefore, we can ultimately conclude that $f(yf(a))\cdot f(y+a)=f(yf(b))\cdot f(y+b)$ so $f(y+a)=f(y+b)$ for any $y\in\mathbb{R}_{>0}.$

The lattter is equivalent to the fact that $f$ is periodic (from a point on) with period $p:=|a-b|>0.$ This cannot hold, because by fixing $x$ (big enough for periodicity to apply) and taking $c\in\mathbb{N}$ big enough such that $c\cdot p+x-f^2(x)>0,$ we reach a contradiction because \[Q(x,x+c\cdot p-f^2(x))\implies f(x)>f(x+c\cdot p).\]Therefore, we cannot have $f(a)=f(b)$ with $a\neq b$ so $f$ is injective, as desired. $\blacksquare$

We finished with the hard part. By combining both claims we firstly infer that $f^2(x)=x$ for all $x\in\mathbb{R}_{>0}.$ Therefore, $P(1,y)$ yields $f(y+1)=f(1)/(y+1)$ so, in other words, $f(x)=f(1)/x$ for all $x>1.$ A couple of other plugins combined with the fact that $f^2(x)=x$ will ultimately imply that $f(x)=f(1)/x$ for all $x.$
This post has been edited 2 times. Last edited by oVlad, Mar 25, 2022, 8:48 AM
Reason: My friend pointed out some typos :P
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Orestis_Lignos
558 posts
#18
Y by
We begin with some Claims ($f^k(x)$ denotes the $k-$th iteration of $f$):

Claim 1: $y>f^2(x)$ implies $f(x)>f(y)$.
Proof: Indeed, if $y>f^2(x)$ we may take $y \rightarrow y-f^2(x)$ in the given assertion, and so $f(x)=f(y)+f(xf(y-f^2(x)))f(x+y-f^2(x))>f(y)$, as desired $\blacksquare$

Claim 2: $f^2(x) \geq x$ for all $x>0$.
Proof: Indeed, suppose that there existed a $x$ such that $x>f^2(x)$. By Claim 1, we would have $f(x)>f(x)$, a contradiction $\blacksquare$

Claim 3: $f^4(x)=f^2(x)$ for all $x>0$.
Proof: Suppose that there existed a $x$ such that $f^4(x) \neq f^2(x)$. By Claim 2 we have $f^4(x)>f^2(x)$, and so by Claim 1 $f(x)>f^5(x)$. However, by Claim 2,

$f(x)>f^5(x) \geq f^3(x) \geq f(x),$

a contradiction $\blacksquare$

Claim 4: $f(x)<\frac{f(1)}{x-1}$ for all $x>1$.
Proof: Indeed, take $x=1$ and $y \rightarrow x-1$ in the given, and show $f(1)>f(f(x-1))f(x) \geq (x-1)f(x)$, which gives the desired result $\blacksquare$

Claim 5: $f$ is injective.
Proof: Suppose there existed $a>b$ such that $f(a)=f(b)$. Then, take $x \rightarrow f^2(x)$ in the given, and so by using Claim 3 we have,

$f^3(x)=f(f^2(x)+y)+f(f^2(x)f(y))f(f^2(x)+y),$

and now if we take $y=a$ and $y=b$ and comparing, we have $f(f^2(x)+a)=f(f^2(x)+b)$ for all $x$. Hence, by taking $y=a$ and $y=b$ in the given and comparing we get $f(x+a)=f(x+b)$.

Thus, $f(x)=f(x+a-b)$ for all $x>a$, and subsequently $f(x)=f(x+k(a-b))$ for all $x>a$ and positive integers $k$.

Now, fix $x$ such that $x>\max \{1,a \}$, and so $f(x+k(a-b))<\frac{f(1)}{x-1+k(a-b)}$, hence $f(x)<\frac{f(1)}{x-1+k(a-b)}$, which is a contradiction for $k$ pretty large. Thus, $f$ must be injective $\blacksquare$

To the problem, by Claims 3 and 5 we have $f^2(x)=x$, hence the given rewrites as $f(x)=f(x+y)(f(xf(y))+1)$. Take $x=1$ here, and so $f(x)=\frac{f(1)}{x}$ for all $x>1$.

Now take a $x>\max \{1, \frac{1}{a} \}$ and $y=\frac{xa}{k}$ with $k \leq 1$. Then, $y=\frac{xa}{k}>\frac{1}{k} \geq 1$, and so we have
$\frac{a}{x}=\frac{a}{x+y}(f(x \cdot \frac{k}{x})+1),$

hence $f(k)=\frac{a}{k}$ for all $k \leq 1$.

Thus, we have $f(x)=\frac{f(1)}{x}$ for all $x$, which works.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
rama1728
801 posts
#19
Y by
Excellent problem! We begin by proving some trivial claims.
Claim 1. \(f(f(x))\geq x\).
Proof. Assume not. Then, \(P(x,x-f(f(x)))\) gives us a contradiction. $\blacksquare$
Claim 2. \(f^4=f^2\).
Proof. I claim that if \(x>f(f(y))\), then \(f(y)>f(x)\). This is simply \(P(y,x-f(f(y)))\). This implies that if \(f(f(x))>f(f(y))\), then \(f(y)>f(x)\) by claim 1. Now, if \(f^4>f^2\), then \(f\le f^3<f\), a contradiction. And from claim 1, we must have \(f^4\ge f^2\), so I conclude that \(f^4=f^2\). $\blacksquare$
Claim 3. \(xf(x)\) is bounded.
Proof. First of all, note that \(P(f^2(x), y)\) and claim 2 gives us that \[\frac{f^3(x)}{f(f^2(x)+y)}=1+f(f^2(x)f(y))\]Therefore, \[\frac{f^3(x)}{f(f^2(x)+f(y))}=1+f(f^2(x)f^2(y))=\frac{f^3(y)}{f(f^2(y)+f(x))}\]And so \[\frac{f^3(x)}{f(f^2(x)+f(y))}=\frac{f^3(y)}{f(f^2(y)+f(x))}\]Now replace \(x\) with \(f^2(x)\) and \(y\) with \(f(y)\). We see that
\[\frac{f^5(x)}{f(f^4(x)+f^2(y))}=\frac{f^4(y)}{f(f^3(y)+f^3(x))}\]and as \(f^4=f^2, f^5=f^3\) we see that \[\frac{f^3(x)}{f(f^3(x)+f^3(y))}=\frac{f^2(y)}{f(f^2(x)+f^2(y))}\]Switching \(x\) and \(y\) and comparing, we see that \(f^2(x)f^3(x)\) is constant. Let this constant be \(c\). Now, \[c=f^2(x)\cdot f^3(x)\geq xf(x)\]and we're done. $\blacksquare$
Claim 4. \(f\) is injective.
Proof. I claim that \(f\) is either injective or periodic. Indeed, let \(f(a)=f(b)\). Let \(\alpha\) be such that \(f^2(\alpha)=\alpha\). Then, \(P(\alpha,a)\) and \(P(\alpha,b)\) give us that \(f(a+\alpha)=f(b+\alpha)\). Now, note that \(\alpha\) can be \(f^2(x)\) for all \(x\), from claim 2. So, we have \(f(f^2(x)+a)=f(f^2(x)+b)\). Now, compare \(P(x,a)\) and \(P(x,b)\) to get \(f(x+a)=f(x+b)\) for all \(x\in\mathbb{R}^+\) and therefore \(f\) is periodic. Now, let the period be \(t\). Notice that \(P(x+t,y)\) and \(P(x,y)\) give us that \(f(x)=f(x+tf(y))\). The thing here is, \(tf(x)\) can be as small and as large as possible. Because from claim \(3\), choose \(x\) really large, this must imply \(f(x)\) really small and from claim 1, choosing \(x\) really large implies that \(f^2(x)\) is really large. We will use this as an advantage. So, we see that \[f(x+u)=f(x+v)\]where \(\lvert u-v\rvert\) is as large as possible. Now, let \(x_0\) be a fixed positive real so that \(v>f(x_0)>u\), note \(u\) is super small and \(v\) is super large. Then, replacing \(x\) with \(f(x_0)-u\) gives us that \[f(f(x_0))=f(f(x_0)+v-u)\]and remember \(f(f(x_0))>x_0\) and \(f(f(x_0)+v-u)<\frac{c}{f(x_0)+v-u}\). Therefore, \[x_0<f(f(x_0))=f(f(x_0)+v-u)<\frac{c}{f(x_0)+v-u}\]implying that \[x_0f(x_0)+x_0(v-u)<c\]But note that we can keep \(u\) the same and make \(v\) as big as we can! From the inequality, we MUST need \(x_0(v-u)<c\) for all possible values for \(v\), but if we make \(v\) tending to infinity, which we of course can, we get a contradiction. This forces \(f\) injective. $\blacksquare$

Now for the final blow, we have that \(f\) is an involution from claim \(1\). And as we got \(f^2\cdot f^3=c\), we see that \(xf(x)=c\), implying that \(f(x)=\frac{c}{x}\) for all positive reals \(x\), which is indeed, a solution.


PS: I just realized how stupid I was, we have \(f(x+kt)=f(x)\) and let \(k\) be a super large positive integer then we are easily done on the injectivity argument. Moreover, we dont even need my claim 3, because once we get \(f\) is injective, then \(f\) is involutive, so \[f(x)=f(x+y)(1+f(xf(y)))\]plug \(x=1\) and so \(f(x)=\frac{c}{x}\) for all \(x>1\). Now let \(y>1\) and \(x>1\). \(x<\frac{y}{c}\). Note all \(t<1\) can be written as \(\frac{cx}{y}\). Then just do \(P(x,y)\) and finish for positive reals at most \(1\).

Oops, a small thing, for the injectivity finish, we do require \(P(1,x-1)\) and then you dont need my claim 3, (else you do need lol) to get \(f(x)<\frac{f(1)}{x-1}\), my bad.

I posted the above solution because that is what I found. Looking back at it, it does not feel too much like an AMO P3, but since I took ~2 hours, I'd rather not comment more on this.
This post has been edited 2 times. Last edited by rama1728, Mar 25, 2022, 11:56 AM
Reason: ;
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
USJL
540 posts
#20 • 2 Y
Y by CantonMathGuy, oVlad
USJL wrote:
Also, the entire motivation behind this FE is...... Galois theory????? It's a long story and "Galois theory" is no longer involved in this version but yeah maybe I'll talk about it later.
The backstory is typed up in the section 3.5 of the document here:
https://artofproblemsolving.com/community/c6h2784600p24469107
This post has been edited 1 time. Last edited by USJL, Mar 25, 2022, 6:38 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
jasperE3
11430 posts
#21
Y by
megarnie wrote:
Claim 1: If $f$ is involution, we are done.
Proof: Suppose $f(f(x))=x$ for all $x$.

$P(1,x): f(1)=f(1+x)+xf(x+1)\implies f(x+1)=\frac{f(1)}{x+1}$, as desired. $\blacksquare$

When I did it, I had to show more than this, since this only gives it for $x>1$.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
megarnie
5658 posts
#22
Y by
jasperE3 wrote:
megarnie wrote:
Claim 1: If $f$ is involution, we are done.
Proof: Suppose $f(f(x))=x$ for all $x$.

$P(1,x): f(1)=f(1+x)+xf(x+1)\implies f(x+1)=\frac{f(1)}{x+1}$, as desired. $\blacksquare$

When I did it, I had to show more than this, since this only gives it for $x>1$.

thanks. fixed.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
jasperE3
11430 posts
#23
Y by
Let $f^k(x)$ denote $f$ iterated $k$ times, and let $P(x,y)$ be the given assertion.

Claim 1:: If $f(x)\le f(y)$, then $y\le f^2(x)$ (similarly if $y>f^2(x)$ then $f(x)>f(y)$)
Both of the statements follow by letting $y>f^2(x)$ and taking:
$P(x,y-f^2(x))\Rightarrow f(x)>f(y)$

Claim 2: $f^4(x)=f^2(x)$
Suppose $f^2(u)<u$ for some $u$. Then:
$P(u,u-f^2(u))\Rightarrow f(uf(u-f^2(u)))f(2u-f^2(u))=0$
which is impossible, so $f^2(x)\ge x$ for all $x$.
Since $f^5(x)\ge f^3(x)\ge f(x)$, we have $f(x)\le f^5(x)$. Then, $f^4(x)\le f^2(x)$ by Claim 1. From $f^2(x)\ge x$, we have $f^4(x)=f^2(x)$.

Claim 3: $f$ is injective
Suppose there are $a<b$ such that $f(a)=f(b)$.
$P(f(f(x)),a),P(f(f(x)),y)\Rightarrow f(f^2(x)+a)=f(f^2(x)+b)$ (using Claim 2)
$P(x,a),P(x,b)\Rightarrow f(x+a)=f(x+b)$
By induction, $f(2a)=f(2a+n(b-a))$ for $n\in\mathbb N$. Take $n$ large enough so that $n>\left|\frac{f^2(2a)-2a}{b-a}\right|+1000$, then $2a+n(b-a)>f^2(2a)$, so by Claim 1, $f(2a)>f(2a+n(b-a))$ which is absurd.

Claim 4: $\boxed{f(x)=\frac ax}$ for $a\in\mathbb R^+$
Let $a=f(1)$. From $f^4(x)=f^2(x)$ we have $f^2(x)=x$, so $P(x,y)$ becomes:
$$f(x)=(1+f(xf(y)))f(x+y).$$Now
$P(1,x)\Rightarrow f(x+1)=\frac a{x+1}$
so $f(x)=\frac ax$ for $x>1$. Now let $y\le1$, and let $x>\max\left\{\frac1y,1-f(y)\right\}+100$. Then:
$P(x,f(y))\Rightarrow f(x)=(1+f(xy))f(x+f(y))\Rightarrow \frac ax=(1+\frac a{xy})\cdot \frac a{x+f(y)}\Rightarrow f(y)=\frac ay$
so the claim is proven. $\frac ax$ is a solution for any positive $a$.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ZETA_in_olympiad
2211 posts
#25 • 2 Y
Y by Mango247, Mango247
How much MOHS would you rate this one? Personally I would rate it 30 MOHS, although I believe 25 isn't that bad either :)
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
GoodMorning
826 posts
#26 • 1 Y
Y by solasky
ZETA_in_olympiad wrote:
How much MOHS would you rate this one? Personally I would rate it 30 MOHS, although I believe 25 isn't that bad either :)

I'd say 30, as this was >>> 2016 USAMO 4
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ZETA_in_olympiad
2211 posts
#27
Y by
GoodMorning wrote:
ZETA_in_olympiad wrote:
How much MOHS would you rate this one? Personally I would rate it 30 MOHS, although I believe 25 isn't that bad either :)

I'd say 30, as this was >>> 2016 USAMO 4

The two problems are different with different solutions, could you explain what you mean?
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
GoodMorning
826 posts
#28 • 2 Y
Y by solasky, megarnie
ZETA_in_olympiad wrote:
GoodMorning wrote:
ZETA_in_olympiad wrote:
How much MOHS would you rate this one? Personally I would rate it 30 MOHS, although I believe 25 isn't that bad either :)

I'd say 30, as this was >>> 2016 USAMO 4

The two problems are different with different solutions, could you explain what you mean?


16 USAMO/4 was a very standard real FE containing no unstandard tricks and which you can make nontrivial progress from trivial substitutions. Whereas for this one I found the jump to partial involution—and only being able to prove injectivity from that point on to be not as motivated.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ZETA_in_olympiad
2211 posts
#29
Y by
GoodMorning wrote:
16 USAMO/4 was a very standard real FE containing no unstandard tricks and which you can make nontrivial progress from trivial substitutions. Whereas for this one I found the jump to partial involution—and only being able to prove injectivity from that point on to be not as motivated.

P1s/P4s are typically standard, and this is P3 which is ought to be tough nevertheless :)
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
pi271828
3389 posts
#30 • 2 Y
Y by MarkBcc168, megarnie
ZETA_in_olympiad wrote:
GoodMorning wrote:
16 USAMO/4 was a very standard real FE containing no unstandard tricks and which you can make nontrivial progress from trivial substitutions. Whereas for this one I found the jump to partial involution—and only being able to prove injectivity from that point on to be not as motivated.

P1s/P4s are typically standard, and this is P3 which is ought to be tough nevertheless :)

he never brought up the difficulty relative to its placement
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
ZETA_in_olympiad
2211 posts
#31
Y by
Let $P(x,y)$ be the assertion and let $f^n(k)$ be the $n$ iterations of $f.$

If $k>f^2(x),$ then $P(x,k-f^2(x))\implies f(x)>f(k).$ From this we have $f^2(x)\geq x\implies f^4(x)\geq f^2(x).$ But comparing with $k\mapsto f^4(x)$ we get $f^4(x)=f^2(x).$

Lemma: $f$ is an injection.
Proof. Assume $\exists u>v: f(u)=f(v).$ Then comparing $P(f^2(x),u)$ and $P(f^2(x),v)$ yields $f(f^2(x)+u)=f(f^2(x)+v).$ Using this compare $P(x,u)$ and $P(x,v)$ to get $f(x+u)=f(x+v)$ but this forces $f$ to periodic, which is easily seen to be false. $\blacksquare$

$P(1,x-1)\implies f(x)=f(1)/x=c/x~\forall x>1.$ But it's not hard to conclude that it holds for all $x\in \mathbb{R}_{>0}.$
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Iora
194 posts
#33
Y by
Not gonna lie, the question was really hard for me and I needed a hint to start, solved anyways.
Claim 1. $f^2(x) \ge x$.
Assume otherwise, then $P(x,x-f^2(x))$ gives $f(x)=f(x)+C$ where $C >0$, hence contraction.
Putting $x= f^2(x)$, $f^4(x) \ge f^2(x) \ge x$
Claim 2. $f^4(x)=f^2(x)$.
If our claim is not true, then $f^4(x)>f^2(x) \ge x$ and $P(x,f^4(x)-f^2(x))$ gives
$$ f(x) = f^5(x) + C$$where $C>0$, which means $f(x)>f^5(x)$. But it is wrong since using Claim 1. we have $f(x) \le f^3(x) \le f^5(x)$, hence $f^2(x)=f^4(x)$
Claim 3. Function is injective
Suppose $f(a)=f(b)$ where $b>a$ . Observing $P(f^2(x),a),P(f^2(x),b)$ we have $$f(f^2(x)+a)=f(f^2(x)+b)$$. Using the original equation, we have $f(x+a)=f(x+b)$. Since $x$ is arbitrary, we can make $b$ arbitrarly bigger such that $x+b > f^2(x+a)$, which implies ( from inequlity $f(x) > f(f^2(x)+y)$) $f(x+b)<f(x+a)$, hence contraction, $a=b$\
Since $f^4(x)=f^2(x)$, by injectivity, $f^2(x)=x$, which implies surjectivity.Then
$$f(x)= f(x+y) ( 1 + f(xf(y)))$$Since plug in $x=1$
$$f(1) = f(1+y)(1+y)$$Let $x=1+y$, then $f(x)= \frac{f(1)}{x}$, which gives answer as $f(x) = \frac{c}{x}$ for constant $c$, which indeed satisfises the f.e
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
jj_ca888
2725 posts
#34 • 1 Y
Y by Mango247
Let $P(x, y)$ denote the assertion.

Note that $P(x, x - f^2(x))$ gives $f(x) = f(x) + \text{something}$, impossible, meaning that $x - f^2(x) \leq 0 \implies f^2(x) \geq x$ for all $x > 0$.

Next, we prove $f$ injective. Suppose there is $0 < a < b$ for which $f(a) = f(b)$. Then, comparing $P(f^2(x), a)$ and $P(f^2(x), b)$ gives\[f(f^2(x) + a) = f(f^2(x) + b).\]Now, comparing $P(x, a)$ and $P(x, b)$ yields $f(x + a) = f(x + b)$, which is amazing because we can write $f(x + a) = f(x + C(b - a))$ for sufficiently large $C$. Now, we can write $f(x + a) > f(f(x + a)^2 + y)$ and make $y$ sufficiently large as well to make $f^2(x + a) + y = x + C(b - a)$ to force $f(x + a) > f(x + b)$, a contradiction. Hence, $a = b$, and $f$ injective.

Now, we prove $f^4(x) = f^2(x)$. Else, $f^4(x) > f^2(x)$, which is what we assume. $P(x, f^4(x) - f^2(x))$ yields $f(x) > f^5(x)$, impossible since $f^5(x) \geq f^3(x) \geq f(x)$. Hence, $f^4(x) = f^2(x)$ and from injectivity $f$ is an involution with $f^2(x) = x$, and is thus bijective.

Now, the FE becomes $f(x) = f(x + y)(1 + xf(y))$. Now, $x = 1$ yields $f(y + 1) = \tfrac{f(1)}{y + 1}$.

This proves $f(x) = \tfrac Cx$ for $x > 1$. For specific $\epsilon < 1$, we can take $P(\epsilon, y)$ for large $y$ to show $f(\epsilon) = \tfrac {C}{\epsilon}$, proving that our final solution set is indeed just $f(x) = \tfrac Cx$.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
CANBANKAN
1302 posts
#36
Y by
Here is a rewrite of my solution in PM#3 that doesn't use injectivity.

The answer is $f(x)\equiv \frac ax$ which clearly works. Let $P(x,y)$ denote the assertion in the question. Define $f^k(x) = f(f^{k-1}(x))$.

Claim 1 $x\le f^2(x)$.

Proof: Assume not, then $P(x,x-f^2(x))$ gives $f(x)=f(x)$+smth nonzero, contradiction.

Corollary 1 $f^3(x)\ge f(x)$ by applying claim 1 on $f(x)$ (as $f(\mathbb{R}^+) \subseteq \mathbb{R}^+$)

Claim 2: If $f^2(x)>f^2(y)$ then $f(x)<f(y)$.

Proof: $P(y,f^2(x)-f^2(y))$ gives $f(y)>f^3(x)\ge f(x)$.

Claim 3: $f^2(x)=f^4(x)$ for all $x\in \mathbb{R}^+$

Proof: Assume not. Then $f^2(x)<f^4(x)$ and $f(x)<f^3(x)$. Let $A=f(f(x))$, then we have $f(x)<f(A)$ and $f(f(x))<f(f(A))$, which contradicts claim 2.

Claim 4: $f(x)<\frac{f(1)}{x-1}$

Proof: $P(1,x-1)$.

Claim 5: $f(x)= f^3(x)$ for all $x\ge 1$

Proof: Let $B=f^2(x)$. We are given $f^2(x)=f^2(B)$ by claim 3.

Assume $f(x)<f(B)$ since $f(x)\le f^3(x)$ by corollary 1. This also means $1\le x<B.$ Then $P(x,y), P(B,y)$ yields$$f(x)=f(f(f(x))+y)+f(x+y)f(xf(y))$$$$f(B)=f(f(f(B))+y)+f(B+y)f(Bf(y))$$
Let $t\to\infty$. Plugging in $y=f(t)$ and subtracting the second equation by the first give $f(B)-f(x)< f(B+f(t))f(Bf(f(t))) <\frac{f(1)}{B-1} \cdot \frac{f(1)}{Bf(f(t))-1} \le \frac{f(1)}{B-1} \cdot \frac{f(1)}{Bt-1}$ which tends to 0 as $t$ tends to infinity.

Claim 6: $f( f^2(x)+f(y)) = f(x+f(y))$ for all $y\ge 1$.

Proof: $P(f(x),y): f(f(x)) = f( f(x)+y ) ( 1 + f(f(x)f(y)))$

$P(f(x), f(f(y))): f(f(x)) = f( f(x)+f(f(y))) (1+f(f(x)f(y)))$

Thus, both quantities are equal to $\frac{f(f(x))}{1+f(f(x)f(y))}$

Claim 7: $f( xf(f(y))) = f( f(f(x)) \cdot f(f(y)))$ for all $x\ge 1$

$P(x,f(y))$: $f(x)=f( f(f(x))+f(y)) + f(x+f(y))f(xf(f(y)))$

$P(f(f(x)),f(y))$: $f(x)= f( f(f(x))+f(y)) + f( f(f(x))+f(y)) f( f(f(x)) f(f(y)))$

Since $f(x+f(y))=f(f(f(x))+f(y))$ it follows that $f(xf(f(y))) = f( f(f(x)) f(f(y)))$ for all $x\ge 1$

Finally, note $Im(f^3)\subseteq Im(f^2)\subseteq Im(f)=Im(f^3)$ (this is over $\mathbb{R}^+$) so they are all the same. In particular, $f( xf(y)) = f( f(f(x)) \cdot f(y))$ for all $x\ge 1$

Now, say $x\ge 1$. Letting $B=f(f(x))$. $P(B,y)-P(x,y)$ give $f(Bf(y))f(B+y)=f(xf(y))f(x+y)$. We just proved $f(Bf(y))=f(xf(y))$. Therefore, $f(B+y)=f(x+y)$. Setting $y=f(f(z))-x$ gives $B=x$, which proves $f(x)=\frac ax$ for $x\ge 1$. The rest is easy.
This post has been edited 6 times. Last edited by CANBANKAN, Feb 15, 2023, 6:19 AM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
YaoAOPS
1579 posts
#37 • 1 Y
Y by asdf334
Denote the assertion with $P(x, y)$.
We have that if $a > f(f(x))$, $f(a) < f(x)$. Equivalently, if $f(a) \ge f(x)$ then $a \ge f(f(x))$ As such, $f(f(x)) \ge x$.

Claim: $f(f(x))$ has a fixed point.
Proof. Note that $f(x) \le f^{3}(x) \le f^{5}(x)$.
Then, if $f^{4}(x) > f(f(x))$ it follows that $f^{5}(x) < f(x)$, contradiction.
Thus, $f(f(x)) = f^{4}(x)$ and $f^{2}(x)$ is a fixed point. $\blacksquare$

Claim: $f$ is strictly decreasing on $[c, \infty)$ for some $c$. Furthermore, $c \le f(f(x))$ for all $x$.
Proof. Let $c$ be the minimal fixed point of $f \circ f$. Then, by $P(c, x)$ it follows that \[ f(c) = f(c + y)(1 + f(cf(y))) \]so $f(c) > f(c + y)$. $\blacksquare$

Claim: $f$ is injective and thus an involution.
Proof. Suppose that $f(a) = f(b)$ for $a \ne b$.
Then, by $P(f(f(x)), a)$ and $P(f(f(x)), b)$ \[ f(f(f(x)) + a) = f(f(f(x)) + b) \]However, $f$ is strictly decreasing on this domain, contradiction.
Since $f^{4}(x) = f(f(x))$ it follows that $f(f(x)) = x$. $\blacksquare$
$P(x, y)$ thus simplifies as \[ f(x) = f(x + y)(1 + f(xf(y))). \]Thus, by $P(1, x)$ \[ \frac{f(1)}{f(1 + x)} = (1 + x) \]so $f(x) = \frac{f(1)}{x}$ for $x \ge 1$. Since $f$ is a involution this holds for all $x$.
It can be seen that for all $f(x) = \frac{c}{x}$, $f$ is a involution and that \[ \frac{c}{x} = \frac{c}{x + y} + \frac{c}{x \cdot \frac{c}{y}} \cdot \frac{c}{x + y} \]works.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
IAmTheHazard
5005 posts
#38 • 1 Y
Y by centslordm
this problem repaired my marriage

The answer is $f(x)=C/x$ only, which clearly work.

First note that
$$f(x)>f(f^2(x)+y),\qquad (\heartsuit),$$so $f^2(x) \geq x$, otherwise we can plug in $y=x-f^2(x)$ and find a contradiction. Let $f(1)=c$; by plugging in $x=1$, we find that
$$c>f(f(y))f(1+y) \geq yf(y+1) \implies f(x)<\frac{1}{x-1}~\forall x>1.$$In particular, $\lim_{x \to \infty} f(x)=0$.

I claim that $f^4(x)=f^2(x)$ for all $x$. Suppose otherwise, so $f^4(x)>f^2(x)$. On the other hand, we also have $f^5(x)\geq f^3(x) \geq f(x)$ for all $x$. Then by $(\heartsuit)$, we get a contradiction by setting $y=f^4(x)-f^2(x)$, because now $f(x)>f^5(x)$.

Suppose that for some $a$, $f^3(a)>f(a)$. Then by replacing $x$ with $f^2(x)$ in the original FE and using the fact that $f^4(x)=f^2(x)$, we obtain
$$f^3(x)=f(f^2(x)+y)(1+f(f(x)^2f(y)))~\forall x.$$Then by comparing $y=f(a)$ and $y=f^3(a)$ we find that $f(f^2(x)+f(a))=f(f^2(x)+f^3(a))$. Then by comparing $y=f(a)$ and $y=f^3(a)$ in the original FE we find that $f(x+f(a))=f(x+f^3(a))$ for all $x$. By iterating this it follows that $f(x+f(a))=f(x+f^3(a))=f(x+(2f^3(a)-f(a)))=\cdots$, and since $f^3(a)>f(a)$ we find an unbounded "chain" of positive reals which produce the same output when $f$ is applied, but this contradicts $\lim_{x\to\infty}f(x)=0$. Thus $f^3(x)=f(x)$ for all $x$. Then by repeating the exact same argument by supposing that $f^2(a)>a$ for some $a$, we find that $f^2(x)=x$ for all $x$, i.e. $x$ is an involution.

With this new information, rewrite the FE as
$$f(x)=f(x+y)(1+f(xf(y))).$$Replace $x$ with $f(x)$ to find that
$$\frac{x}{f(f(x)+y)}=1+f(f(x)f(y)).$$Swapping $x$ and $y$ then implies that
$$\frac{x}{y}=\frac{f(f(x)+y)}{f(f(y)+x)}.$$But then simultaneously replacing $x$ and $y$ with $f(x)$ and $f(y)$ respectively yields
$$\frac{f(x)}{f(y)}=\frac{f(f(y)+x)}{f(f(x)+y)}=\frac{y}{x},$$hence $xf(x)$ is constant, i.e. $f(x)=C/x$ for some $x$, which is the desired result. $\blacksquare$
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
joshualiu315
2535 posts
#39 • 1 Y
Y by dolphinday
The answer is $f(x) = \boxed{\tfrac{c}{x}}$ for $c>0$. This is shown to work, so we will prove it is the only solution. Denote the given assertion as $P(x,y)$ and at any point in this proof, $f^n(x)$ denotes $n$ iterations of $f$.

Consider the following assertion:

\[Q(a,b): \ f(a) \ge f(b) \implies f(f(b)) \ge a. \]
To see that this is true, we will consider the contrapositive. If $a > f(f(b))$, then $P(b,a-f(f(b)))$ yields

\[f(b) = f(a) + \text{blah} > f(a),\]
which finishes the proof.


Claim 1: $f^2(x) = f^4(x)$

Proof: Notice that $Q(a,a)$ yields

\[f^2(a) \ge a.\]
Then, plugging in $a = f(x)$ and $a=f^3(x)$ gives

\[f^5(x) \ge f^3(x) \ge f(x).\]
This means $Q(f^4(x), x)$ holds; in other words,

\[f^2(x) \ge f^4(x).\]
But, we know that $f^4(x) \ge f^2(x)$ by plugging in $a = f(x)$. Therefore, $f^2(x) = f^4(x)$ as desired. $\square$


Claim 2: $f$ is injective

Proof: Suppose that $f(a) = f(b)$ for two reals $a>b$. Comparing $P(f^2(x),a)$ and $P(f^2(x),b)$ yields

\[f(f^4(x)+a) + f(f^2(x)f(a)) f(f^2(x)+a) = f(f^4(x)+b) + f(f^2(x)f(b)) f(f^2(x)+b)\]\[\implies f(f^2(x)+a) \left[1 + f(f^2(x)f(a)) \right] = f(f^2(x)+b) \left[1 + f(f^2(x)f(b)) \right].\]
This implies that $f(f^2(x)+a) = f(f^2(x)+b)$. Thus, $f$ is periodic.

However, taking an arbitrarily large value of $k$ such that $f(x) = f(x+k)$ and $f^2(x) < x+k$ leads to a contradiction due to $Q$. $\square$


From injectivity, Claim 1 becomes $f(f(x)) = x$. Plugging this into $P$ yields

\[f(x) = f(x+y)+f(xf(y))f(x+y) = f(x+y) \left(1+f(xf(y)) \right).\]
Substituting in $x=1$ yields

\[f(1) = f(y+1) (1+f(f(y)))\]\[ \implies f(y+1) = \frac{f(1)}{y+1},\]
which means $f(x) = \tfrac{c}{x}$, as desired.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
EthanWYX2009
889 posts
#40
Y by
Not very difficult after proving $f$ is injective.
https://cdn.aops.com/images/b/3/2/b323610c412ce717de658cd27639333cdcb66d5b.jpg
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
OronSH
1774 posts
#41 • 4 Y
Y by Alex-131, Sedro, megarnie, ihatemath123
We claim $f(x)=\frac cx$ for constant $c$.

First notice $f(f(x))\ge x$, as otherwise $P(x,x-f(f(x)))$ gives a contradiction.

Next, if $f(f(x))<f(y)$, then $P(x,f(y)-f(f(x)))$ implies $f(x)>f(f(y))\ge y$.

Now we claim $f(f(x))=f(f(f(f(x))))$. If $f(f(x))<f(f(f(f(x))))$, then $f(x)>f(f(f(x)))$, but we have $f(f(f(x)))\ge f(x)$.

Next by taking $P(f(f(x)),f(y))$ and $P(f(f(x)),f(f(f(y))))$ gives $f(f(f(x))+f(y))=f(f(f(x))+f(f(f(y))))$.

Now $P(x,f(y))$ and $P(x,f(f(f(y))))$ give $f(x+f(f(f(y))))=f(x+f(y))$. Thus either $f(f(f(y)))=f(y)$ for all $y$, or $f$ is eventually periodic with period $f(f(f(y)))-f(y)=p>0$.

However this is impossible since it would imply $f(f(x))=f(f(x+kp))\ge x+kp$ for $x$ large enough and all positive integers $k$.

Now take $P(x,f(f(y)))$. We get \begin{align*}f(x)&=f(f(f(x))+f(f(y)))+f(xf(y))f(x+f(f(y)))\\&=f(f(f(x))+f(f(y)))+f(xf(y))f(y)-f(xf(y))f(yf(x))f(x+y).\end{align*}Thus by symmetry we have \[\frac{f(x)}{f(y)}=\frac{1+f(yf(x))}{1+f(xf(y))}.\]
Now if $f(a)=f(b)$ then setting $x=a,x=b$ we get $f(af(y))=f(bf(y))$ for all $y$. Now $P(a,y),P(b,y)$ give $f(a+y)=f(b+y)$, so $a=b$ or $f$ is eventually periodic, which is impossible.

Thus $f$ is injective, so $f(f(x))=f(f(f(f(x))))$ implies $x=f(f(x))$.

Finally, setting $x,y=f(x),f(y)$ in $\frac{f(x)}{f(y)}=\frac{1+f(yf(x))}{1+f(xf(y))}$ gives $xf(x)=yf(y)$, so $f(x)=\frac cx$ for some constant $c$.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
awesomeming327.
1780 posts
#42 • 1 Y
Y by OronSH
The answer is $f(x)=\tfrac{c}{x}$ for all $c\in \mathbb{R}_{>0}$, which clearly works.

Claim: $f(f(x))\ge x$.
Assume $f(f(x_0))<x_0$ then $P(x_0,x_0-f(f(x_0)))$ gives
\[f(x_0f(y))f(x_0+y)=0\]which is a contradiction because the range of $f$ does not include $0$.
Claim: $f^4(x)=f^2(x)$.
We already know that $f^4(x)\ge f^2(x)$. Suppose $f^4(x_0)>f^2(x_0)$ for some $x_0$ then $P(x_0,f^4(x_0)-f^2(x_0))$ gives
\[f(x_0)=f^5(x_0)+f(x_0f(y))f(x_0+y)>f^5(x_0)\]but $f^5(x_0)\ge f^3(x_0)\ge f(x_0)$, so that's a contradiction.
Claim: $f$ is injective.
Taking $P(f^2(x),y)$ gives
\[f^3(x)=f(f^2(x)+y)(1+f(f^2(x)f(y)))\]Suppose $f(y_1)=f(y_2)$ with $y_1\neq y_2$ then $f(f^2(x)+y_1)=f(f^2(x)+y_2)$ for all $x$. Now take regularly $P(x,y_1)$ and $P(x,y_2)$ gives
\[f(x+y_1)=\frac{f(x)-f(f^2(x)+y_1)}{f(xf(y_1))}=\frac{f(x)-f(f^2(x)+y_2)}{f(xf(y_2))}=f(x+y_2)\]which implies $f$ is eventually periodic. Let $k$ be its fundamental period, then for all $x$, $P(x,k)$ gives
\[f(x)=f^3(x)+f(xf(y))f(x)\]but $f^3(x)\ge f(x)$, so this is impossible.
Since $f^4(x)=f^2(x)$, we have $f^2(x)=x$, so our equation simplifies to
\[f(x)=f(x+y)(1+f(xf(y)))\]Let $Q(x,y)$ denote the above assertion. Comparing $Q(f(x),y)$ and $Q(f(y),x)$ gives
\[\frac{x}{f(f(x)+y)}=\frac{y}{f(f(y)+x)}\]while comparing $Q(x,f(y))$ and $Q(y,f(x))$ gives
\[\frac{f(x)}{f(x+f(y))}=\frac{f(y)}{f(y+f(x))}\]These two together imply that $xf(x)=yf(y)$ which implies $f(x)=c/x$ for some $c$. We are done.
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
cj13609517288
1939 posts
#43 • 2 Y
Y by OronSH, megarnie
The answer is $\boxed{f(x)=\frac{c}{x}}$ for any positive real $c$, which works because
\[\frac{c}{x}=\frac{c}{x+y}+\frac{c}{xc/y}\cdot\frac{c}{x+y}.\]
Note that $x\ne f(f(x))+y$, so $f(f(x))\ge x$. Then
\[f(f(f(x))+y)<f(x)\le f(f(f(x))).\]Therefore, if $u=f(f(x))$ for some $x$, then $f(u+y)<f(u)$ for all $y$.

Claim. $f^2(x)=f^4(x)$ for all $x$, where the superscript denotes repeated applications.
Proof. Suppose not. Then $f^4(x)=f(f(f^2(x)))>f^2(x)$, so
\[f^3(x)\le f(f(f^3(x)))=f(f^4(x))<f(f^2(x)),\]contradiction. $\blacksquare$

Claim. $f$ is injective.
Proof. Set $f(a)=f(b)$. Then fix some $x$, and set $u=f(f(x))$. Comparing $P(u,y)$ for $y\in\{a,b\}$ gives that $f(u+a)=f(u+b)$. Then comparing $P(x,y)$ for $y\in\{a,b\}$ gives that $f(x+a)=f(x+b)$. Thus $f$ is eventually periodic, say with period $T$ and starting at $x>x_0$. Then pick some $x_1>x_0$ and note that
\[x_1+Tn\le f(f(x_1+Tn))=f(f(x_1))\]for all positive integers $n$, absurd. $\blacksquare$

Now $f(f(x))=x$ for all $x$, so
\[f(x)=f(x+y)(1+f(xf(y))).\qquad(\clubsuit)\]Set $c=f(1)$. Plugging in $x=1$ gives
\[c=(1+y)f(1+y)\Longrightarrow f(y)=\frac{c}{y}\;\;\;\;\forall y>1.\]Now fix some $y<1$ and take $x$ to be very large in $(\clubsuit)$. Then
\[\frac{c}{x}=\frac{c}{x+y}\cdot\left(1+f(xf(y))\right)\Longrightarrow
\frac{y}{x}=\frac{c}{xf(y)}\Longrightarrow f(y)=\frac{c}{y}.\]So we have proven $f(x)=\frac{c}{x}$ for all $x$, as desired. $\blacksquare$
This post has been edited 1 time. Last edited by cj13609517288, Mar 6, 2025, 9:25 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
Ilikeminecraft
726 posts
#44
Y by
Claim: $f(a)\leq f(b) \implies f(f(a)) \geq b$
Proof: AFTSOC, $f(f(a)) < b.$ Take $x = a, y = b - f(f(a)) > 0,$ and we get $f(a) = f(b) + f(\text{blah})f(\text{blah}) > f(b),$ contradiction.

Let the inequality be $Q(x, y).$
Claim: $f^4(x) = f^2(x)$
Proof: Take $Q(t, t)$ to get $f^2(t)\geq t.$ Plug in $t = f^2(x)$ and we get $f^4(x)\geq f^2(x).$

Furthermore, $f(x) \leq f^3(x)\leq f^5(x).$ Hence, we can take $Q(f^4(x), x)$ to get $f^4(x) \leq f^2(x).$ This finishes our claim.

Claim: $f$ is injective
Proof: Assume $f(a) = f(b).$ Take $Q(a, b)$ to get $f(f(a)) \geq b,$ and take $Q(b, a)$ to get $f(f(b)) \geq a.$ Then, take $(f^2(x), a), (f^2(x), b)$ in the original equation to get $f^3(x) = f(f^2(x) + b) + f(f^2(x)f(b))f(f^2(x) + b) = f(f^2(x) + a) + f(f^2(x)f(a))f(f^2(x) + a).$ Thus, $f(f^2(x) + b) = f(f^2(x) + a).$

Take $(x, a)$ in the given fe now to get $f(x + a) = f(x + b).$ Hence, $f$ has period $b - a$. Take $Q(1, 1 + n(b - a))$ and we get $f(f(1)) \geq 1 + n(b - a),$ which is obviously a contradiction for some sufficiently large $n.$

Injectivity implies $f$ is an involution.

Hence, $f(x) = f(x + y)(1 + f(xf(y))).$ Pick $x = 1$ and we get $f(1 + y) = \frac{f(1)}{1 + y}.$ Hence, if $f(1) = c,$ we get $f(x) = \frac cx$ for $x > 1.$ Since $f$ is an involution, we have $f(\frac cx) = x = \frac c{\frac cx},$ and by increasing $x$ as required, we get $f\equiv\frac cx.$
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
peace09
5455 posts
#45 • 4 Y
Y by dolphinday, GrantStar, OronSH, ihatemath123
Solved with Jason Lee (@akliu) in half an hour: the answer is $f\equiv\tfrac{c}{x}$, which clearly works.

Immediately $f(x)>f(f^2(x)+y)$; varying $y$ gives $f(x)>f(k)$ for all $k>f^2(x)$. In particular we cannot have $k:=x$, meaning $x\le f^2(x)$. Then $f^2(x)\le f^4(x)$; if $f^2(x)<f^4(x)$, letting $k:=f^4(x)$ produces $f(x)>f^5(x)$, which directly contradicts $f(x)\le f^3(x)\le f^5(x)$. So $f^2(x)=f^4(x)$.

Claim. $f$ is injective.
Proof. Suppose $f(a)=f(b)$. We use the matching arguments trick: the $f(f^2(x)+y)$ and $f(x+y)$ terms in the FE are eerily similar, and in fact $f^2(x)=f^4(x)$ ensures we can match them exactly by setting $x:=f^2(x)$ to attain
\begin{align*}
f^3(x)&=f(f^2(x)+y)+f(f^2(x)f(y))f(f^2(x)+y)\\
&=f(f^2(x)+y)(1+f(f^2(x)f(y))).
\end{align*}Setting $y:=a,b$ gives $f(f^2(x)+a)=f(f^2(x)+b)$, and plugging back into the FE reduces to $f(x+a)=f(x+b)$. If toward contradiction $a\neq b$, letting $k:=x+n|a-b|$ for sufficiently large $n$ yields $f(x)>f(x+n|a-b|)$, bad. $\square$

Hence $x=f^2(x)$; plugging back into the FE results in
\[f(x)=f(x+y)(1+f(xf(y)).\]Setting $x:=1$ we see $f(y+1)=\tfrac{c}{y+1}$ for all $y>0$, and since $f$ is an involution we in turn have $f(x)=\tfrac{c}{x}$ for $x\not\in(c,1)$. Setting $y:=1$ we have $f(x)=\tfrac{c+cf(cx)}{1+x}$ for all $x>0$, recovering the solution set. $\blacksquare$
This post has been edited 7 times. Last edited by peace09, Jun 2, 2025, 11:08 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
pi271828
3389 posts
#46 • 18 Y
Y by ostriches88, speedyfalcon, akliu, megarnie, Schintalpati, GrantStar, Sid-darth-vater, imagien_bad, sus_rbo, KevinYang2.71, mathnerd_101, Danieliu, eggfierice, akIiu, MatthewLiu, majix0418, OronSH, EpicBird08
peace09 wrote:
Solved with pi271828 in half an hour: the answer is $f\equiv\tfrac{c}{x}$, which clearly works.

Immediately $f(x)>f(f^2(x)+y)$; varying $y$ gives $f(x)>f(k)$ for all $k>f^2(x)$. In particular we cannot have $k:=x$, meaning $x\le f^2(x)$. Then $f^2(x)\le f^4(x)$; if $f^2(x)<f^4(x)$, letting $k:=f^4(x)$ produces $f(x)>f^5(x)$, which directly contradicts $f(x)\le f^3(x)\le f^5(x)$. So $f^2(x)=f^4(x)$.

Claim. $f$ is injective.
Proof. Suppose $f(a)=f(b)$. We use the matching arguments trick: the $f(f^2(x)+y)$ and $f(x+y)$ terms in the FE are eerily similar, and in fact $f^2(x)=f^4(x)$ ensures we can match them exactly by setting $x:=f^2(x)$ to attain
\begin{align*}
f^3(x)&=f(f^2(x)+y)+f(f^2(x)f(y))f(f^2(x)+y)\\
&=f(f^2(x)+y)(1+f(f^2(x)f(y))).
\end{align*}Setting $y:=a,b$ gives $f(f^2(x)+a)=f(f^2(x)+b)$, and plugging back into the FE reduces to $f(x+a)=f(x+b)$. If toward contradiction $a\neq b$, letting $k:=x+n|a-b|$ for sufficiently large $n$ yields $f(x)>f(x+n|a-b|)$, bad. $\square$

Hence $x=f^2(x)$; plugging back into the FE results in
\[f(x)=f(x+y)(1+f(xf(y)).\]Setting $x:=1$ we see $f(y+1)=\tfrac{c}{y+1}$ for all $y>0$, and since $f$ is an involution we in turn have $f(x)=\tfrac{c}{x}$ for $x\not\in(c,1)$. Setting $y:=1$ we have $f(x)=\tfrac{c+cf(cx)}{1+x}$ for all $x>0$, recovering the solution set. $\blacksquare$

Bro italicized the only thing he did... (those who know)
This post has been edited 1 time. Last edited by pi271828, Jun 2, 2025, 8:24 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
akliu
1803 posts
#47 • 6 Y
Y by megarnie, Amkan2022, GrantStar, pi271828, OronSH, J55406
where's my credit alex you bum....
This post has been edited 1 time. Last edited by akliu, Jun 2, 2025, 11:14 PM
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
peace09
5455 posts
#48
Y by
Fixed group solve credits (those who know)
Z K Y
The post below has been deleted. Click to close.
This post has been deleted. Click here to see post.
akliu
1803 posts
#49 • 2 Y
Y by peace09, megarnie
peace09 wrote:
Fixed group solve credits (those who know)

Thank you, Alex Liu.
Z K Y
N Quick Reply
G
H
=
a