RM/TRM: take dS_L_dp_cap from time discretization
The derivative dS_L_dp_cap is taken from time discretization for the Jacobian as well, and corrected for p_cap_dot_ip=0. Three tests failed after these changes:
- RM quasinewton LARGE
- RM masslumping small
Results for last time step can be found here:
Significant changes can be found only for RM quasinewton LARGE. For Liakopoulos, it was enough to loosen the vtkdiff criteria for the pressure a bit. After the changes, the STDOUT was smaller (305K) compared to before (352K). For the other tests, changes seem to be marginal, however the output sizes and thereby iterations increased:
- masslumping: before: 448K after: 487K
- Liakoupolus: before: 192K after: 286K
I also checked the normal Richardflow-2D small test, where the number of iterations decreased. So, the number of iterations does not seem to be affected systematically in one way or the other. The total time for all 33 tests stayed about the same on my machine (~140 s).
- Feature description was added to the changelog
- Tests covering your feature were added?
- Any new feature or behavior change was documented?