A Bridge Between Dubovitskiĭ–Federer Theorems and the Coarea Formula
Piotr Hajlasz,Mikhail V. Korobkov,Jan Kristensen
DOI: https://doi.org/10.1016/j.jfa.2016.10.031
IF: 1.891
2016-01-01
Journal of Functional Analysis
Abstract:The Morse-Sard theorem requires that a mapping v: R-n -> R-m is of class C-k, k > max(n - m, 0). In 1957 Dubovitskii generalized this result by proving that almost all level sets for a C-k mapping have H-s-negligible intersection with its critical set, where s = max(n - m - k + 1,0). Here the critical set, or m-critical set is defined as Z(v,m) = {x is an element of R-n : rank del v(x) < m}. Another generalisation was obtained independently by Dubovitskii and Federer in 1966, namely for C-k mappings v : R-n -> R-d and integers m <= d they proved that the set of m-critical values v(Z(v,m)) is H-qo-negligible for q(o), = m - 1 + n-m+1/k. They also established the sharpness of these results within the C-k category.Here we prove that Dubovitskii's theorem can be generalized to the case of continuous mappings of the Sobolev-Lorentz class W-p,1(k)(R-n, R-d), p = n/k (this is the minimal integrability assumption that guarantees the continuity of mappings). In this situation the mappings need not to be everywhere differentiable and in order to handle the set of nondifferentiability points, we establish for such mappings an analog of the Luzin N-property with respect to lower dimensional Hausdorff content. Finally, we formulate and prove a bridge theorem that includes all the above results as particular cases. As a limiting case in this bridge theorem we also establish a new coarea type formula: if E degrees subset of {x is an element of R-n : rank del v(x) <= m}, thenintegral(E) J(m)v(x) dx = integral(Rd) Hn-m (E boolean AND v(-1) (y)) dH(m)(y).The mapping v is R-d-valued, with arbitrary d, and the formula is obtained without any restrictions on the image v(R-n) (such as m-rectifiability or sigma-finiteness with respect to the m-Hausdorff measure). These last results are new also for smooth mappings, but are presented here in the general Sobolev context. The proofs of the results are based on our previous joint papers with J. Bourgain (2013, 2015). (C) 2016 Elsevier Inc. All rights reserved.