As of revision 56, two convergence tests are failing:
pipe_unstructured_stationary() in pytests/full_system/stokes/test_stokes.py
test_sinusoidal_oseen_adaptive_two_step_cg() in pytests/full_system/ins/test_ins.py
However, all other convergence tests are passing and after looking into the current implementation of these tests it might be the case that they are too strict. All of the convergence tests both test for converge of the solver (residual errors are below specified error tolerances) but also test the rate at which the residual errors decrease per nonlinear iteration.
The approach to testing convergence needs to be examined and, potentially, all of the convergence pytests changed to simple confirm convergence after some reasonable maximum number of nonlinear iterations.
As of revision 56, two convergence tests are failing:
pipe_unstructured_stationary() in pytests/full_system/stokes/test_stokes.py
test_sinusoidal_oseen_adaptive_two_step_cg() in pytests/full_system/ins/test_ins.py
However, all other convergence tests are passing and after looking into the current implementation of these tests it might be the case that they are too strict. All of the convergence tests both test for converge of the solver (residual errors are below specified error tolerances) but also test the rate at which the residual errors decrease per nonlinear iteration.
The approach to testing convergence needs to be examined and, potentially, all of the convergence pytests changed to simple confirm convergence after some reasonable maximum number of nonlinear iterations.