English

Uniform Sobolev Resolvent Estimates for the Laplace-Beltrami Operator on Compact Manifolds

Analysis of PDEs 2013-02-26 v2 Classical Analysis and ODEs

Abstract

In this paper we continue the study on the resolvent estimates of the Laplace-Beltrami operator Δg\Delta_g on a compact manifolds MM with dimension n3n\geq3. On the Sobolev line 1/p1/q=2/n1/p-1/q=2/n we can prove that the resolvent (Δg+ζ)1(\Delta_g+\zeta)^{-1} is uniformly bounded from LpL^p to LqL^q when (p,q)(p,q) are within the admissible range p2(n+1)/(n+3)p\leq2(n+1)/(n+3) and q2(n+1)/(n1)q\geq2(n+1)/(n-1) and ζ\zeta is outside a parabola opening to the right and a small disk centered at the origin. This naturally generalizes the previous results in \cite{Kenig} and \cite{bssy} which addressed only the special case when p=2n/(n+2),q=2n/(n2)p=2n/(n+2), q=2n/(n-2). Using the shrinking spectral estimates between LpL^p and LqL^q we also show that when (p,q)(p,q) are within the interior of the admissible range, one can obtain a logarithmic improvement over the parabolic region for resolvent estimates on manifolds equipped with Riemannian metric of non-positive sectional curvature, and a power improvement depending on the exponent (p,q)(p,q) for flat torus. The latter therefore partially improves Shen's work in \cite{Shen} on the LpL2L^p\to L^2 uniform resolvent estimates on the torus. Similar to the case as proved in \cite{bssy} when (p,q)=(2n/(n+2),2n/(n2))(p,q)=(2n/(n+2),2n/(n-2)), the parabolic region is also optimal over the round sphere SnS^n when (p,q)(p,q) are now in the admissible range. However, we may ask if the admissible range is sharp in the sense that it is the only possible range on the Sobolev line for which a compact manifold can have uniform resolvent estimate for ζ\zeta being ouside a parabola.

Keywords

Cite

@article{arxiv.1209.5689,
  title  = {Uniform Sobolev Resolvent Estimates for the Laplace-Beltrami Operator on Compact Manifolds},
  author = {Peng Shao and Xiaohua Yao},
  journal= {arXiv preprint arXiv:1209.5689},
  year   = {2013}
}

Comments

A few details revised