Harnack inequalities for log-Sobolev functions and estimates of log-Sobolev constants (Q1807219): Difference between revisions

Consider \(L=\Delta+\nabla V\) for some \(V\in C^{\infty}(M)\), where \(M\) is a \(d\)-dimensional connected compact Riemannian manifold, and denote by \(\mu\) the probability \(d\mu=Z^{-1}\exp(V(x))dx\) (\(Z=\int\exp(V(x))dx\)). The log-Sobolev constant for \(L\) is \[ \alpha=\inf_{f\in{\mathcal F}}\frac{2\mu(|\nabla f|^{2})}{\mu(f^{2}\log{f^{2}})} \] where \({\mathcal F}=\{f\in C^{1}(M):\mu(f^{2})=1,f\neq\text{ cst.}\}\). A function in \({\mathcal F}\) is called a log-Sobolev function if it achieves the constant \(\alpha\). A log-Sobolev function is a solution to \(Lf=-\frac{\alpha}{2}f\log{f^{2}}\). The author proves a dimension-free Harnack inequality for log-Sobolev functions. The proof is based on the maximum principle [see also \textit{F. R. K. Chung} and \textit{S.-T. Yau}, Math. Res. Lett. 3, No.~6, 793-812 (1996; Zbl 0880.58026)]. Then some lower bound estimates of the log-Sobolev constant are presented by using the spectral gap inequality and the coupling method. This paper recovers and improves some results from \textit{F. R. K. Chung} and \textit{S.-T. Yau.}.