January 2025
M T W T F S S
 12345
6789101112
13141516171819
20212223242526
2728293031  

Categories:

Tags:

On IMO2017 P2

I did IMO 2017 P2 with my friends yesterday. Now I would like to share my thoughts on this problem.

Problem: Let \mathbb{R} be the set of real numbers. Determine all functions f:\mathbb{R}\rightarrow \mathbb{R} such that, for any real numbers x and y,

    \[   f(f(x)f(y))+f(x+y)=f(xy)\]

The initial step must be taking certain values, testing its properties and then start guessing the solution. Well, obviously, the constant function f(x)=0 is a solution to the problem.

Now we can try to take x=y=0, giving us

(1)   \begin{equation*}   f(f(0)\cdot f(0)))+f(0)=f(0)\end{equation*}

and

(2)   \begin{equation*}   f(f(0)f(0)))=0\end{equation*}

Now if we take x=x, y=0, we could get

(3)   \begin{equation*}   f(f(x)f(0)))+f(x)=f(0)\end{equation*}

f(0) appears on both side, and it is quite hard to analyze it. What if we already know the value of f(0)? Suppose that f(0)=0, then f(0)+f(x)=f(0), so f(x)=0, giving us the trivial solution. Now suppose that f(0)\neq 0, notice that if we replace f with -f, then -f is also a solution. That means we can only consider the case where f(0)>0.

Now we are stuck. Maybe we should try to get more informations. One strategy we often use involving solving functional equations is cancelation. Let x+y=xy, we can replace y=\dfrac{x}{x-1}. Plugging in, we get

(4)   \begin{equation*}   f(f(x)f(\frac{x}{x-1}))=0\end{equation*}

This result could be pretty useful, as we can use it to analyze the zeros of f(x). We know that there must have at least one zero. Now suppose that for some f(a), f(a)=0, then f\left(f(a)f(\frac{a}{a-1})\right)=f(0)>0, which is a contradiction. That means if we have a zero, then it must be 1. Therefore, f(1)=0, and f(x)f(\frac{x}{x-1})=1 whenever n\neq 1.

Now we could more or less take a guess. It is impossible that f(x) has a degree greater than one, and it is very likely that it is a linear function, which is f(x)=1-x. We proceed with this conjecture.

Having f(1)=0, we can plug x=1 into (4). We derive

(5)   \begin{equation*} f(0)^2 = 1\end{equation*}

This gives us f(0)=1. So we could try more values. Taking x=x, y=1, we get

(6)   \begin{equation*}   f(f(x)f(1))+f(x+1)=f(x)\end{equation*}

so

(7)   \begin{equation*}   f(x+1)=f(x)-1\end{equation*}

We can also take x=x, y=0, we get

(8)   \begin{equation*}   f(f(x))+f(x)=f(0)=1\end{equation*}

so

(9)   \begin{equation*}   f(f(x))=1-f(x)\end{equation*}

Viewing this composite function, we can use the traditional method.

(10)   \begin{equation*}    f(1-f(x))=f(f(f(x))=1-f(f(x))=f(x)\end{equation*}

The solution could be very similar to our conjecture. If 1-f(x)=x, then f(x)=1-x. That means if we can prove injection, then we are done!

This could be a hard part. We go back to our problem

(11)   \begin{equation*}    f(f(x)f(y))=f(xy)-f(x+y)\end{equation*}

Suppose that f(a)=f(b), we need to pick specific x and y to give some useful information. Seeking relation to zero, we want f(xy)-f(x+y)=1. So let xy=a+1 and x+y=b. This is achievable, since by our previous result, f(x+1)=f(x)-1. If we want the determinant b^2-4(a+1)\geq 0, we can shift a and b so that b is large enough. Finally, we get

(12)   \begin{equation*}    f(f(x)f(y))=f(a+1)-f(b)=1\end{equation*}

So f(x)f(y)=0\rightarrow f(x)=0 or f(y)=0. Therefore, x=1 or y=1, giving us a=b, and we complete the proof.

One response to “On IMO2017 P2”

  1. Ioshift Avatar
    Ioshift

    Cool

Leave a Reply

Your email address will not be published. Required fields are marked *