[petsc-users] help needed for -snes_mf_operator

[Dominik Szczerba]

>> Running with the options:
>>
>> -ksp_type bcgs -pc_type jacobi -ksp_rtol 1e-4 -ksp_max_it 1000
>> -ksp_converged_use_initial_residual_norm -ksp_norm_type
>> unpreconditioned -ksp_monitor_true_residual -ksp_converged_reason
>> -ksp_view -snes_monitor -snes_converged_reason -snes_ls_monitor
>> -snes_view -snes_type test -snes_test_display
>>
>> I get the output:
>>
>> Testing hand-coded Jacobian, if the ratio is
>> O(1.e-8), the hand-coded Jacobian is probably correct.
>>
>> and the procedure stagnates, calling FormJacobian only once, and then
>> FormFunction infinitely, with my own computation of ||F|| indicating
>> stagnation.
>
>   It is not calling it infinitely. Run with the -snes_type test option for a small problem, with say 10 unknowns to start.
>
>   Barry

Thanks for a useful hint. The smallest problem I was able to generate
has around 1000 unknowns but it finally outputs something. I got some
matrix, with the biggest absolute entry order 0.1 the smallest
absolute entry order 1e-14, average values seem to be order 1e-8 or
so. End line of the output says: Norm of matrix ratio 0.579314
difference 10.9951 (constant state 1.0)

Dominik