A Matrix-Free Algorithm for the Large-Scale Constrained Trust-Region Subproblem
Anthony J. Kearsley
A new matrix-free algorithm for the solution of linear inequality constrained, large-scale trust-region sub-problems is presented. The matrix-free nature of the algorithm eliminates the need for any matrix factorizations and only requires inner products between vectors and rows/columns of matrices. Numerical results that demonstrate the viability of the approach are included.
constrained quadratic optimization, lanczos method, regularization, trust region
A Matrix-Free Algorithm for the Large-Scale Constrained Trust-Region Subproblem, Optimization Methods and Software, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=50695
(Accessed June 5, 2023)