Font Size: a A A

Constrained Optimization Problem Of Trust Region Methods

Posted on:2002-03-15Degree:MasterType:Thesis
Country:ChinaCandidate:Z S YuFull Text:PDF
GTID:2190360032957421Subject:Operational Research and Cybernetics
Abstract/Summary:PDF Full Text Request
This paper explores the convergence of trust region methods for constrained minimization of differentiable function.In the first part, two trust region algorithms for linearly constrained optimization problem are proposed 1)Combining quasi-Newton and trust region method for linearly equality constrained optimization is given , its global convergence and super-linear rate is proved. 2) Interior-point trust region algorithm for box constrained problem and its global convergence. In the second part , a non-monotone trust region algorithm for non-linearly equality constrained optimization problem is proposed and its global convergence and super-linear rate is proved.
Keywords/Search Tags:trust region algorithm, non-monotone trust region algorithm, interior-point trust region algorithm, quasi-Newton algorithm, global convergence, super-linear rate
PDF Full Text Request
Related items