Second order subdifferentials constructed using integral convolutions smoothing
The aim of this paper is to introduce a new Hessian that is constructed from the local Hessian information via an integral convolution smoothing. It is then compared to several other kinds of known subhessians. The main question is how effective are these constructions in capturing the essential second-order information associated with the function. Sections 2, 3 and 4 of this paper give a detailed survey on the tools used from variational analysis, on mollifier subjets, and on para-concavity. The main results are contained in Sections 4 and 5 and give exhaustive answers to the motivating problems of the paper. One of the important observations is that the application of the infimal convolution approximation prior to the integral convolution smoothing could remove the severe restrictions on the class of para-concave functions. For the entire collection see [Zbl 1055.90002].
- Second-order subgradients of convex integral functionals
- Sub-hessians, super-hessians and conjugation
- Epi-convergence properties of smoothing by infimal convolution
- scientific article; zbMATH DE number 599156
- Normal and mixed partial second-order subdifferentials in variational analysis and optimization
This page was built for publication: Second order subdifferentials constructed using integral convolutions smoothing
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4667371)
