SNESSetFromOptions#

Sets various SNES and KSP parameters from user options.

Synopsis#

#include "petscsnes.h"  
PetscErrorCode SNESSetFromOptions(SNES snes)

Collective

Input Parameter#

snes - the SNES context

Options Database Keys#

-snes_type - newtonls, newtontr, ngmres, ncg, nrichardson, qn, vi, fas, SNESType for complete list
-snes_stol - convergence tolerance in terms of the norm of the change in the solution between steps
-snes_atol - absolute tolerance of residual norm
-snes_rtol - relative decrease in tolerance norm from initial
-snes_divergence_tolerance - if the residual goes above divtol*rnorm0, exit with divergence
-snes_force_iteration - force SNESSolve() to take at least one iteration
-snes_max_it <max_it> - maximum number of iterations
-snes_max_funcs <max_funcs> - maximum number of function evaluations
-snes_max_fail <max_fail> - maximum number of line search failures allowed before stopping, default is none
-snes_max_linear_solve_fail - number of linear solver failures before SNESSolve() stops
-snes_lag_preconditioner - how often preconditioner is rebuilt (use -1 to never rebuild)
-snes_lag_preconditioner_persists <true,false> - retains the -snes_lag_preconditioner information across multiple SNESSolve()
-snes_lag_jacobian - how often Jacobian is rebuilt (use -1 to never rebuild)
-snes_lag_jacobian_persists <true,false> - retains the -snes_lag_jacobian information across multiple SNESSolve()
-snes_tr_tol - trust region tolerance
-snes_convergence_test - <default,skip,correct_pressure> convergence test in nonlinear solver. default SNESConvergedDefault(). skip SNESConvergedSkip() means continue iterating until max_it or some other criterion is reached, saving expense of convergence test. correct_pressure SNESConvergedCorrectPressure() has special handling of a pressure null space.
-snes_monitor [ascii][:filename][:viewer format] - prints residual norm at each iteration. if no filename given prints to stdout
-snes_monitor_solution [ascii binary draw][:filename][:viewer format] - plots solution at each iteration
-snes_monitor_residual [ascii binary draw][:filename][:viewer format] - plots residual (not its norm) at each iteration
-snes_monitor_solution_update [ascii binary draw][:filename][:viewer format] - plots update to solution at each iteration
-snes_monitor_lg_residualnorm - plots residual norm at each iteration
-snes_monitor_lg_range - plots residual norm at each iteration
-snes_monitor_pause_final - Pauses all monitor drawing after the solver ends
-snes_fd - use finite differences to compute Jacobian; very slow, only for testing
-snes_fd_color - use finite differences with coloring to compute Jacobian
-snes_mf_ksp_monitor - if using matrix-free multiply then print h at each KSP iteration
-snes_converged_reason - print the reason for convergence/divergence after each solve
-npc_snes_type - the SNES type to use as a nonlinear preconditioner
-snes_test_jacobian - compare the user provided Jacobian with one computed via finite differences to check for errors. If a threshold is given, display only those entries whose difference is greater than the threshold.
-snes_test_jacobian_view - display the user provided Jacobian, the finite difference Jacobian and the difference between them to help users detect the location of errors in the user provided Jacobian.

Options Database Keys for Eisenstat-Walker method#

-snes_ksp_ew - use Eisenstat-Walker method for determining linear system convergence
-snes_ksp_ew_version ver - version of Eisenstat-Walker method
-snes_ksp_ew_rtol0 - Sets rtol0
-snes_ksp_ew_rtolmax - Sets rtolmax
-snes_ksp_ew_gamma - Sets gamma
-snes_ksp_ew_alpha - Sets alpha
-snes_ksp_ew_alpha2 - Sets alpha2
-snes_ksp_ew_threshold - Sets threshold

Notes#

To see all options, run your program with the -help option or consult the users manual

SNES supports three approaches for computing (approximate) Jacobians: user provided via SNESSetJacobian(), matrix free, and computing explicitly with finite differences and coloring using MatFDColoring. It is also possible to use automatic differentiation and the MatFDColoring object.