Confidence Regions for Global Optima in Nonlinear Programming.
TEXAS A AND M UNIV COLLEGE STATION INST OF STATISTICS
Pagination or Media Count:
The report is concerned with developing new statistical techniques for nonlinear optimization including nonconvex optimization. The approach is a statistical one and provides an upper confidence limit for the global maximum of a mathematical function gx of a vector x in a multi-dimensional feasible space, say S. Specifically the report develops the statistical techniques for determining these confidence limits as well as algorithms implementing the techniques and computer programs executing the algorithms. Modified author abstract
- Operations Research