Model Justified Search Algorithms for Scheduling Under Uncertainty
Abstract
We have identified clear trade-offs in algorithm design for the scheduling under uncertainty problem that relate to the sources of the uncertainty. We also identified plateaus as a significant barrier to superb performance of local search on scheduling and have studied several canonical discrete optimization problems to discover and model the nature of plateaus. From this, we have developed lower and upper bound predictive models of plateau size in a significant optimization problem: MAXSAT. We also develop new theoretical results on the nature of plateaus in Elementary Landscapes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 25, 2008
- Accession Number
- ADA492118
Entities
People
- Adele Howe
- L. D. Whitley
Organizations
- Colorado State University