Hello
I have a general question about the effect of time step in the computed results. I am solving a 2D coupled nonlinear problem which is similar to Convection-Diffusion Example . My code is very long and could not be broken down to a simple example. I am just presenting the solver part of the code:
problem = NonlinearVariationalProblem(F, z, bcs, J)
solver = NonlinearVariationalSolver(problem)
solver.parameters['newton_solver']['convergence_criterion'] = 'incremental'
solver.parameters['newton_solver']['linear_solver'] = 'mumps'
My problem converges well when the size of time step is 1 second. For each time step it takes almost 3 iterations to converge. Here is an example:
Newton iteration 0: r (abs) = 6.272e+00 (tol = 1.000e-10) r (rel) = 1.000e+00 (tol = 1.000e-09)
Newton iteration 1: r (abs) = 7.740e-06 (tol = 1.000e-10) r (rel) = 1.234e-06 (tol = 1.000e-09)
Newton iteration 2: r (abs) = 1.890e-12 (tol = 1.000e-10) r (rel) = 3.013e-13 (tol = 1.000e-09)
Newton solver finished in 3 iterations and 3 linear solver iterations.
When I increase the size of time step, it seems like it does not have a significant effect on the solution. For example here is a graph explaining what I am trying to say:
It looks a bit strange to me because when I increase the size of time step from 1 second to 20 seconds I expect it affects the results (Maybe I am wrong!). In addition the problem still converges well in the same 3 iteration in each time step.
I appreciate if somebody could clarify this behavior and explain if this is normal or not.
Thanks