petsc-3.9.4 2018-09-11
Report Typos and Errors

TAOLMVM

Limited Memory Variable Metric method is a quasi-Newton optimization solver for unconstrained minimization. It solves the Newton step Hkdk = - gk using an approximation Bk in place of Hk, where Bk is composed using the BFGS update formula. A More-Thuente line search is then used to computed the steplength in the dk direction

Options Database Keys

-tao_lmm_vectors - number of vectors to use for approximation
-tao_lmm_scale_type - "none","scalar","broyden"
-tao_lmm_limit_type - "none","average","relative","absolute"
-tao_lmm_rescale_type - "none","scalar","gl"
-tao_lmm_limit_mu - mu limiting factor
-tao_lmm_limit_nu - nu limiting factor
-tao_lmm_delta_min - minimum delta value
-tao_lmm_delta_max - maximum delta value
-tao_lmm_broyden_phi - phi factor for Broyden scaling
-tao_lmm_scalar_alpha - alpha factor for scalar scaling
-tao_lmm_rescale_alpha - alpha factor for rescaling diagonal
-tao_lmm_rescale_beta - beta factor for rescaling diagonal
-tao_lmm_scalar_history - amount of history for scalar scaling
-tao_lmm_rescale_history - amount of history for rescaling diagonal
-tao_lmm_eps - rejection tolerance
-tao_lmm_recycle - enable recycling LMVM updates between TaoSolve() calls

Level

beginner

Location

src/tao/unconstrained/impls/lmvm/lmvm.c

Examples

src/tao/unconstrained/examples/tutorials/eptorsion1.c.html
src/tao/unconstrained/examples/tutorials/rosenbrock1.c.html
src/tao/unconstrained/examples/tutorials/rosenbrock1f.F90.html

Index of all Tao routines
Table of Contents for all manual pages
Index of all manual pages