Difference between revisions of "2023 USAMO Problems/Problem 2"

Line 121: Line 121:
 
</cmath>
 
</cmath>
 
doesn't necessarily mean the derivative is constant either. It only implies the derivative is periodic with period <math>f(1)</math>
 
doesn't necessarily mean the derivative is constant either. It only implies the derivative is periodic with period <math>f(1)</math>
 +
 
spencerD.
 
spencerD.

Revision as of 12:12, 11 May 2023

Problem 2

Let $\mathbb{R}^{+}$ be the set of positive real numbers. Find all functions $f:\mathbb{R}^{+}\rightarrow\mathbb{R}^{+}$ such that, for all $x, y \in \mathbb{R}^{+}$,\[f(xy + f(x)) = xf(y) + 2\]

Solution

First, let us plug in some special points; specifically, plugging in $x=0$ and $x=1$, respectively:

\begin{align}     f(f(0)) &= 2 \\     f(y + f(1)) &= f(y) + 2 \end{align}

Next, let us find the derivative of this function. First, with (2), we isolate $f(y)$ one one side

\begin{align*}    f(y) = f(y + f(1)) - 2 \end{align*}

and then take the derivative:

\begin{align*}     \dfrac{\mathrm{d}f}{\mathrm{d}y}     &= \dfrac{\mathrm{d}f}{\mathrm{d}y}\left[f(y + f(1)) - 2\right] \\     &= \dfrac{\mathrm{d}f}{\mathrm{d}y}\left[f(y + f(1))\right] - \dfrac{\mathrm{d}f}{\mathrm{d}y}\left[2\right] \\     &= f'(y + f(1))\cdot\dfrac{\mathrm{d}f}{\mathrm{d}y}\left[y + f(1)\right] \\     &= f'(y + f(1))\cdot(1)\\     f'(y) &= f'(y + f(1))\\ \end{align*}

With the derivative, we see that the input to the function does not matter: it will return the same result regardless of input, assuming that $f(1) \neq 0$. We know it is not zero because if it was, then (2) would become $f(y) = f(y) + 2$, implying that $0 = 2$, which is not true.

Therefore, the function $f'$ must be a constant, and $f$ must be a linear equation or a constant. We know it is not a constant because if it was, the problem could be reduced to the following:

\begin{align*}     f(xy + f(x)) &= xf(y) + 2 \\     f &= xf + 2 \\     f - xf &= 2 \\     f(1-x) &= 2 \\     f &= \dfrac{2}{1-x} \\ \end{align*}

where $f$ is the constant from $f(x)$. As we see, $f$ would depend on $x$, making it not a constant function. Thus, $f(x)$ must be linear, meaning we can model it like so:

\begin{align*}     f(x) = ax + b \end{align*}

Via (1), we get the following:

\begin{align*}     f(f(0)) &= 2 \\     a(a(0) + b) + b &= 2 \\     ab + b &= 2 \end{align*}

And via (2),

\begin{align*}     f(y + f(1)) &= f(y) + 2 \\     a(y + a(1) + b) + b &= ay + b + 2 \\     ay + a^2 + ab + b &= ay + b + 2 \\     a^2 + ab &= 2 \\ \end{align*}

Setting these equations equal to each other,

\begin{align*}     ab + b &= a^2 + ab \\     b &= a^2 \\ \end{align*}

Therefore,

\begin{align*}     ab + b &= 2 \\     a^3 + a^2 &= 2 \\ \end{align*}

There are three solutions to this equation: $a = 1$, $a = -1 + i$, and $a = -1 - i$. Knowing that $b = a^2$, the respective $b$ values are $b = 1$, $b = -2i$, and $b = 2i$. Thus, $f(x)$ could be the following:

\begin{align*}     f(x) &= x + 1 \\     f(x) &= x(-1 + i) - 2i \\     f(x) &= x(-1 - i) + 2i \\ \end{align*}

Because only the first function maps strictly to positive real numbers, the only solution that works is $f(x) = x + 1$. $\square$

~cogsandsquigs

This solution is heavily flawed:

1. $f(0)$ doesn't exist.

2. The function isn't necessarily continuous, let alone differentiable.

3. Even if it was differentiable, the equation \begin{align*} f'(y) &= f'(y + f(1))\\ \end{align*} doesn't necessarily mean the derivative is constant either. It only implies the derivative is periodic with period $f(1)$

spencerD.