Exploiting Explicit and Implicit Structure in Complex Optimization Problems
Abstract
To solve many practical complex optimization problems often results in having to minimize a nonsmooth function f. The graph of such a function of n variables appears V-shaped in some directions while in directions orthogonal to these the graph of f's closely related "U-Lagrangian" is U-shaped. To be efficient a minimization algorithm needs to approximate the correct "VU-space decomposition" and any existing second order information on the U-subspace via some kind of corresponding combination of polyhedral (V-shaped) and quadratic (U-shaped) approximation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 31, 2011
- Accession Number
- ADA563754
Entities
People
- Robert B. Mifflin
Organizations
- Washington State University