English

The Dubovitski\u{\i}-Sard Theorem in Sobolev Spaces

Classical Analysis and ODEs 2015-06-02 v1

Abstract

The Sard theorem from 1942 requires that a mapping f:RnRmf:\mathbb{R}^n \to \mathbb{R}^m is of class CkC^k, k>max(nm,0)k > \max (n-m,0). In 1957 Duvovitski\u{\i} generalized Sard's theorem to the case of CkC^k mappings for all kk. Namely he proved that, for almost all yRmy\in \mathbb{R}^m, H(Cff1(y))=0\mathcal{H}^{\ell}(C_f \cap f^{-1}(y))=0 where =max(nmk+1,0)\ell = \max(n-m-k+1,0), H{\mathcal H}^{\ell} denotes the Hausdorff measure, and CfC_f is the set of critical points of ff. In 2001 De Pascale proved that the Sard theorem holds true for Sobolev mappings of the class Wlock,p(Rn,Rm)W_{\rm loc}^{k,p}(\mathbb{R}^n,\mathbb{R}^m), k>max(nm,0)k>\max(n-m,0) and p>np>n. We will show that also Dubovitski\u{\i}'s theorem can be generalized to the case of Wlock,p(Rn,Rm)W_{\rm loc}^{k,p}(\mathbb{R}^n,\mathbb{R}^m) mappings for all kNk\in\mathbb{N} and p>np>n.

Keywords

Cite

@article{arxiv.1506.00025,
  title  = {The Dubovitski\u{\i}-Sard Theorem in Sobolev Spaces},
  author = {Piotr Hajłasz and Scott Zimmerman},
  journal= {arXiv preprint arXiv:1506.00025},
  year   = {2015}
}