FCI Tests and options for testing #423
Conversation
… option for parallel divergence
Damp gradients of velocity rather than gradients of momentum.
Damp gradients of velocity rather than gradients of momentum.
Also, there is no need to add a symlink
Working on tests
totork
changed the title
|
I think it might be wise to keep it like this now. It is already quite alot of changes. Changed to be ready @bendudson @dschwoerer |
| continue; | ||
| } | ||
|
|
||
| //const Field3D tau = 1. / get<Field3D>(species["collision_frequency"]); |
There was a problem hiding this comment.
| //const Field3D tau = 1. / get<Field3D>(species["collision_frequency"]); |
|
|
||
| // Calculate ion collision times | ||
| const Field3D tau = 1. / floor(get<Field3D>(species["collision_frequency"]), 1e-10); | ||
| //const Field3D tau = 1. / floor(get<Field3D>(species["collision_frequency"]), 1e-10); |
There was a problem hiding this comment.
| //const Field3D tau = 1. / floor(get<Field3D>(species["collision_frequency"]), 1e-10); |
| ddt(NV) -= AA * FV::Div_par_fvv<hermes::Limiter>(Nlim, V, fastest_wave, fix_momentum_boundary_flux); | ||
|
|
||
|
|
||
| if (isMMS) { |
There was a problem hiding this comment.
This feels a bit wrong, to use a different method in MMS testing than in production. The Div_par_fvv operator converges for non-FCI, so the issue must be in the FCI branch of Div_par_fvv.
There was a problem hiding this comment.
If I use the standard operator, I am getting first order convergence. I presume that this is due to the dissipation of the operator? @ZedThree had a similar experience with testing this operator. I could also use this operator and set the "accepted" mark to first order instead of second order.
3 participants
This PR is aiming to add the following things:
- Both tests have their required files are tar files added.
A crucial physics change is the use of only the self-collision time for each species conduction/viscosity. This should be in line with Braginskii? The proper treatment shall be pulled from master, but it is getting refactored now