A linear method of finite penalties with regulated ℝà-descent

A linear method of finite penalties with regulated ℝà-descent

0.00 Avg rating—0 Votes
Article ID: iaor1990698
Country: United Kingdom
Volume: 28
Start Page Number: 1
End Page Number: 7
Publication Date: Apr 1988
Journal: USSR Computational Mathematics and Mathematical Physics
Authors:
Abstract:

A finite penalty method, based on a linear approximation of a non-linear conditional optimization problem and using •-active constraints with a regulated parjameter •, is considered. The method is shown to be non-locally convergent and estimates for the rate of convergence, valid for the gradient search method in unconditional optimization, are established.

Reviews

Required fields are marked *. Your email address will not be published.