Penalty Functions and Duality in Stochastic Programming via phi-Divergence Functionals.
Abstract
This paper considers stochastically constrained nonlinear programming problems, A penalty type method is suggested as a deterministic surrogate. The penalty is constructed in terms of a 'distance' function between random variables, given a term of the phi-divergence functional (a generalization of the relative entropy). A duality theory is developed in which a general relation between phi-divergence and utility functions is revealed, via the conjugate transform, and a new type of certainty equivalent concept emerges. Additional keywords: computations; vector analysis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1984
- Accession Number
- ADA153985
Entities
People
- A. Ben-tal
- Marc Teboulle
Organizations
- University of Texas at Austin