Jets, generalised convexity, proximal normality and differences of functions (Q4259677)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1328200
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Jets, generalised convexity, proximal normality and differences of functions |
scientific article; zbMATH DE number 1328200 |
Statements
Jets, generalised convexity, proximal normality and differences of functions (English)
0 references
21 June 2000
0 references
generalized convexity
0 references
generalized conjugation
0 references
subjets
0 references
superjets
0 references
proximal subdifferentials
0 references
0 references
0 references
0.765986979007721
0 references
0.761345624923706
0 references
0.7612516283988953
0 references
0.7534514665603638
0 references
A function \(f:\mathbb{R}^n\to\overline{\mathbb{R}}\) is said to possess a subjet at \(x\in\text{dom }f\) if the set \[ D^{2,-}f(x)= \{(\nabla\varphi(x), \nabla^2\varphi(x)): f-\varphi\text{ has a local minimum at \(x\) with }\varphi\in C^2\}; \] if the set \[ D^{2,+}f(x)= \{(\nabla f(x), \nabla^2\varphi(x)): f-\varphi\text{ has a local maximum at \(x\) with }\varphi\in C^2\} \] is nonempty then \(f\) is said to possess a superjet at \(x\). The authors relate subjets and superjets, as well as proximal subdifferentials, to the notion of \(\Phi\)-convexity in the sense of \textit{S. Dolecki} and \textit{S. Kurcyusz} [SIAM J. Control Optimization 16, 277-300 (1978; Zbl 0397.46013)]; the class of functions considered is \[ \Phi_2= \{a-(c/2)\|\cdot-y\|^2: (a,c,y)\in \mathbb{R}\times \mathbb{R}_+\times \mathbb{R}^n\}. \] For \(\varepsilon> 0\), they define the \(\varepsilon\)-proximal subdifferential of \(f\) at \(x\) as \(\partial^\varepsilon_{\Phi_2}f(x)= \partial_{\Phi_2}(f+ \varepsilon\|\cdot-x\|)(x)\), \(\partial_{\Phi_2}\) denoting \(\Phi_2\)-subdifferential, and show that \[ \nabla\partial^\varepsilon_{\Phi_2}f(x)\subseteq \partial^-_\varepsilon f(x)\subseteq \nabla\partial^{\varepsilon+ \delta}_{\Phi_2} f(x), \] where \(\delta>0\), \(\nabla\partial^\varepsilon_{\Phi_2}(x)= \{\nabla\varphi(x): \varphi\in \partial^\varepsilon_{\Phi_2}(x)\}\), and \(\partial^-_\varepsilon\) stands for Ioffe \(\varepsilon\)-subdifferential, given by \[ \partial^-_\varepsilon f(x)= \{z:\langle z,y\rangle\leq d^- f(x;y)+ \varepsilon\|y\|\text{ for all }y\}, \] with \(d^-f(x;y)= \liminf_{t\downarrow 0, y'\to y}(1/t)(f(x+ ty')- f(x))\). The paper also contains a generalization of the Phelps-Bronsted-Rockafellar theorem in terms of \(\Phi_2\)-subgradients, and characterizations of the proximal and Ioffe-Mordukhovich subdifferentials of the differences of functions.
0 references