2
$\begingroup$

Background:

Let $\mathbb{S}^2_+$ be the hemisphere. Then we know that for $f:\mathbb{S}^2_+\to \mathbb{R}$ satisfying (when written in coordinates) $\int_{0}^{2\pi}\int_{0}^{\pi/2}f(r,\theta)\sin(r)dr d\theta = 0$ we have $$\int_{0}^{2\pi}\int_{0}^{\pi/2} |f|^2 \sin(r)drd\theta \leq \frac{1}{2}\int_{0}^{2\pi}\int_{0}^{\pi/2} |\nabla f|^2 \sin(r)drd\theta.$$

Here the best constant $1/2$ can be computed by simply recalling the first eigenvalue on the hemisphere (which is equal to two) along with fact that the poincare constant is $\frac{1}{\lambda_1(\mathbb{S}^2_+)}.$

Question:

I am interested in understanding the case when the test functions in the above inequality are of the form $f(r,\theta)=f(r)$. In particular if we work with $\int_{0}^{\pi/2} f(r)\sin(r)dr=0$ then what is the best constant $C>0$ in the inequality

$$\int_{0}^{\pi/2} |f|^2 \sin(r)dr \leq C \int_{0}^{\pi/2} |\partial_r f|^2 \sin(r)dr?$$

$\endgroup$
5
  • 2
    $\begingroup$ (i) "we know that [...]" -- How/from where do we know that? (ii) Shouldn't $\int_{-\pi/2}^{\pi/2}$ be replaced everywhere by $\int_0^{\pi/2}$? $\endgroup$ Commented Feb 8 at 18:52
  • $\begingroup$ I agree with Iosif that $r$ should run on $[0, \pi/2]$. Also, if you show some details on the proof of the first statement, it will probably shed light on the second. It may well be that a maximizer to the first inequality is radial (i.e. it depends on $r$ only), in which case the second follows. $\endgroup$ Commented Feb 8 at 18:58
  • $\begingroup$ @IosifPinelis (i) mathoverflow.net/questions/295683/… might be relevant (ii) you are right, thanks for pointing that out! $\endgroup$
    – Student
    Commented Feb 8 at 23:02
  • $\begingroup$ @GiuseppeNegro I am guessing the extremizers are eigenfunctions on the hemisphere, which are the coordinate functions. Writing them co-ordinates, $x_1=\sin(r)cos(\theta),x_2=\sin(r)\sin(\theta).$ Then $\int \sin(r)^3 = 2/3$ while $\int \cos(r)^2 \sin(r) = 1/3$ so I am guessing that the constant has to be at least $2$? $\endgroup$
    – Student
    Commented Feb 8 at 23:13
  • $\begingroup$ Oh wait $\int \sin(r)^2\neq 0,$ maybe I am missing something here.... $\endgroup$
    – Student
    Commented Feb 8 at 23:20

0

You must log in to answer this question.