Article
Keywords:
normed space; uniform convexity; closed set; metric projection; $l^p$-space; Fréchet differential; Lipschitz condition
Summary:
Let $F$ be a closed subset of $\mathbb R^n$ and let $P(x) $ denote the metric projection (closest point mapping) of $x\in \mathbb R^n$ onto $F$ in $l^p$-norm. A classical result of Asplund states that $P$ is (Fréchet) differentiable almost everywhere (a.e.) in $\mathbb R^n$ in the Euclidean case $p=2$. We consider the case $2<p<\infty $ and prove that the $i$th component $P_i(x)$ of $P(x)$ is differentiable a.e.\ if $P_i(x)\neq x_i$ and satisfies Hölder condition of order $1/(p-1)$ if $P_i(x)=x_i$.
References:
[4] Federer, H.:
Geometric Measure Theory. Die Grundlehren der mathematischen Wissenschaften in Einzeldarstellungen 153, Springer, Berlin (1969).
MR 0257325 |
Zbl 0176.00801
[7] Kruskal, J. B.:
Two convex counterexamples: A discontinuous envelope function and a non-differentiable nearest point mapping. Proc. Am. Math. Soc. 23 (1969), 697-703.
DOI 10.2307/2036613 |
MR 0259752 |
Zbl 0184.47401
[10] Rešetnyak, Ju. G.:
Generalized derivatives and differentiability almost everywhere. Mat. Sb., N. Ser. 75(117) (1968), 323-334 Russian.
MR 0225159 |
Zbl 0165.47202
[12] Zajíček, L.:
On differentiation of metric projections in finite dimensional Banach spaces. Czech. Math. J. 33 (1983), 325-336.
MR 0718916 |
Zbl 0551.41048