Valadier-like formulas for the supremum function I
R. Correa, A. Hantoute, M.A. L\'opez

TL;DR
This paper extends Valadier's formulas for the subdifferential of the supremum of convex functions by removing continuity assumptions, providing a more general and simplified characterization applicable in broader contexts.
Contribution
It generalizes Valadier's original subdifferential formula for supremum functions by eliminating the continuity requirement, offering a more comprehensive and simplified characterization.
Findings
Removed the continuity assumption in Valadier's formula.
Derived a general subdifferential formula involving nearby data functions.
Provided a simplified Valadier formula when the supremum is continuous at some point.
Abstract
We generalize and improve the original characterization given by Valadier [18, Theorem 1] of the subdifferential of the pointwise supremum of convex functions, involving the subdifferentials of the data functions at nearby points. We remove the continuity assumption made in that work and obtain a general formula for such a subdiferential. In particular, when the supremum is continuous at some point of its domain, but not necessarily at the reference point, we get a simpler version which gives rise to the Valadier formula. Our starting result is the characterization given in [11, Theorem 4], which uses the epsilon-subdifferential at the reference point.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsOptimization and Variational Analysis · Numerical methods in inverse problems · Nonlinear Partial Differential Equations
