Dear All,
I would like to control the maximal number of iteration steps/absolute tolerance for the GMRES algorithm, but I couldn’t find the correct options.
Thanks a lot for information,
Alexei
Are you using PETSc?
It seems that on this particular machine load “PETSc” fails so this is the builtin (?) fgmres implementation,
set(A, solver=GMRES)
Then, I do not know.
How to do it with PETSc?
-ksp_max_it XY -ksp_rtol alpha -ksp_atol beta
. See also KSPSetTolerances — PETSc 3.21.1 documentation and KSPSetFromOptions — PETSc 3.21.1 documentation.
1 Like
Thanks! Is there a way to display/output the number of iterations GMRES actually needed to converge?
-ksp_converged_reason
, it’s in the documentation I sent, so I suggest you read those links…