2
$\begingroup$

Given a locally Lipschitz continuous function $f: \mathbb{R}^{n} \rightarrow \mathbb{R}$ and closed set $$\Omega =\left\lbrace x \in\mathbb{R}^n \ |\ f(x) \leq 0 \right\rbrace$$ such that f is semi-analytic function and analytic inside $\mathbb{R}^{n} \setminus\Omega$ and $\text{Int }{\Omega}$. The Clarke's subdifferential set is the set given by $$ \partial f (x) = {\text{cl co}}\left\lbrace \lim_{k\in\mathbb{N}} \nabla f (x^{k}) \ |\ \text{whenever } x^{k} \rightarrow x^{*} \text{ and } \lim_{k\in\mathbb{N}} \nabla f (x^{k}) \text{ exist} \right\rbrace $$ for all $x \in \mathbb{R}^n.$ My question is about the context in which this multivalued function satisfies a weak notion of locally Lipschitz continuity for multifunctions; by that I mean that for all convex compact set $K$ there exist a $L>0$ such that for all $x\in K$ there exist a $\delta > 0 $ such that for all $y,z \in B(x,\delta) \cap K $ it holds $$ d(\partial f (y), \partial f (z)) = \inf_{g\in \partial f (y)} \inf_{g' \in \partial f (z) } \|g - g'\| < L \|y-z\|. $$ Note that when the function is locally Fréchet differentiable on an open set containing $\Omega$ with continuous second derivative, this is true. It seems to be true for all functions sufficiently regular. It might be true for piecewise smooth functions with sufficiently regular set of discontinuity of the derivative.

Is there a good context in which this is true?

$\endgroup$
8
  • $\begingroup$ You can suppose that $f$ is smooth, if it helps to get rid of the smoothness of f and just having to worry about the boundary of $\Omega.$ "Usually", its not true: the convex function $f(x) = \sqrt{x^3}$ does not satisfy this. $\endgroup$ Apr 10, 2021 at 17:35
  • $\begingroup$ Can you clarify a couple of points? What does $\mathrm{cl} \, \mathrm{co}$ refer to? What role does $\Omega$ play - it seems like you could just consider the question on an open domain $U \subset \mathbf{R}^n$? $\endgroup$
    – Leo Moos
    Apr 13, 2021 at 11:23
  • $\begingroup$ $\text{cl co}$ is the closure of the convex hull of a set: en.wikipedia.org/wiki/Convex_hull#Definitions. Since the convex hull of a compact set is, again, compact, the closure can be omitted. $\endgroup$ Apr 13, 2021 at 21:05
  • $\begingroup$ I have strengthened the hypotheses on $f$ and on $\Omega$ now $\endgroup$ Apr 13, 2021 at 21:38
  • $\begingroup$ It looks that the paper called Lipschitz properties of semi-analytic sets by Adam Parusinski might help with my question. If someone is fluent in algebraic geometry, then it might help with my question. I bet it is an interesting paper. =) $\endgroup$ Apr 13, 2021 at 22:21

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.