2012/12/06 by Jennifer B. Erway, Erway, Jennifer B., Roummel F. Marcia +1
Computer Science · Mathematics · #Advanced Optimization Algorithms Research #FOS: Mathematics #G.4 #Matrix Theory and Algorithms #Numerical Analysis (math.NA) #Numerical methods for differential equations #Optimization and Control (math.OC) #Statistical and numerical algorithms
paper · pdf · doi:10.48550/arxiv.1212.1525
openalex publication_date 2012/12/06 · openalex created_date 2022/09/02 · openalex updated_date 2026/07/28
A MATLAB implementation of the More-Sorensen sequential (MSS) method is\npresented. The MSS method computes the minimizer of a quadratic function\ndefined by a limited-memory BFGS matrix subject to a two-norm trust-region\nconstraint. This solver is an adaptation of the More-Sorensen direct method\ninto an L-BFGS setting for large-scale optimization. The MSS method makes use\nof a recently proposed stable fast direct method for solving large shifted BFGS\nsystems of equations [13, 12] and is able to compute solutions to any\nuser-defined accuracy. This MATLAB implementation is a matrix-free iterative\nmethod for large-scale optimization. Numerical experiments on the CUTEr [3,\n16]) suggest that using the MSS method as a trust-region subproblem solver can\nrequire significantly fewer function and gradient evaluations needed by a\ntrust-region method as compared with the Steihaug-Toint method.\n