0
$\begingroup$

I'm trying to prove the function $f:\mathbb{R}^2 \to \mathbb{R}$ defined as $$f(x,y)=(x^2+y^2)e^{-(x^2+y^2)}$$ attains a maximum at every point of the unit circle. The determinant of the hessian matrix at those points is zero so I wrote $f$ in polar coordinates and tryed to prove there is no $r \in \mathbb{R}$ satisfying $$r^2e^{-r^2}>e^{-1}$$ which is equivalent to $$e^{r^2}<er^2$$ I don't know how to prove it, any suggestion is welcome.

Thanks

$\endgroup$
  • $\begingroup$ Your statement is false for $0\leq r<1$ (and you don't need to consider $r<0$ since the domain is restricted due to the coordinate transformation). $\endgroup$ – LinAlg Jan 18 '17 at 18:32
  • 1
    $\begingroup$ Recall: $e^z\geq 1+z$ for $z\in\mathbb{R}$. Use this with $z=r^2-1$. $\endgroup$ – yurnero Jan 18 '17 at 18:49
0
$\begingroup$

This can be proven more easily. Consider $$f(z) = z e^{-z} (z\geq 0) \quad f'(z) = (1-z)e^{-z}$$ The derivative is $0$ only for $z=1$. Since $f'(z) > 0 $ for $z<1$ and $f'(z) < 0$ for $z>1$, $z=1$ corresponds to a unique maximum.

$\endgroup$
  • $\begingroup$ Of course, didn't realize I had to think it as a one variable extrema problem. Thanks! $\endgroup$ – la flaca Jan 18 '17 at 18:39
1
$\begingroup$

The function $g(r) = e^r$ is strictly convex, therefore its graph lies above each tangent line: $$ e^r = g(r) \ge g(1) + g'(1)(r-1) = er $$ with equality only for $r=1$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.