Let $(M^n,g)$ be a closed Riemannian manifold. Define $$\lambda(g)=\inf\{\mathcal{F}(g,f),\;0<f\in C^{\infty}(M),\; \int_Mf^2\;d\nu=1\},$$ where $$\mathcal{F}(g,f)=\int_M\left(|\nabla f|^2+ af^2\log f + bRf^2\right)\;d\nu,$$ and $R$ is the scalar curvature of manifold.

Can one be confident that there is a $0<f_0\in C^{\infty}(M)$ with $\int_Mf_0^2\;d\nu=1$, such that $\lambda(g)=\mathcal{F}(g,f_0)$?

  • 1
    $\begingroup$ Looks like a job for the direct method. However, the functional is not convex in $f$ due to the term $f^2\log(f)$, so there is some work to do. Another problem may be the "open" constraint $f>0$. Both problems would not be there if there would be $f\log(f)$ instead of $f^2\log(f)$. Since I do not work on manifolds, I have no idea how to formulate the respective Banach spaces (should be similar to some Sobolev or Orlicz-Sobolev spaces…). $\endgroup$ – Dirk Jan 10 at 12:58
  • 1
    $\begingroup$ Does your manifold have finite volume? Otherwise if you choose $f<\epsilon$ everywhere we have $\int f^2 log(f)< log(\epsilon) \int f^2= log(\epsilon)$ and $F(g,f)$ can get arbitrarily small. $\endgroup$ – user100927 Jan 14 at 12:57
  • $\begingroup$ Yes, the volume is finite. $\endgroup$ – user162551 Jan 14 at 21:03

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.