Skip to content
This repository has been archived by the owner on Jul 14, 2022. It is now read-only.

add the kripke #57

Open
wants to merge 216 commits into
base: develop
Choose a base branch
from
Open

add the kripke #57

wants to merge 216 commits into from

Conversation

mclarsen
Copy link
Collaborator

No description provided.

@mclarsen
Copy link
Collaborator Author

@starintheuniverse Here you go.

@mclarsen
Copy link
Collaborator Author

mclarsen commented Oct 31, 2020

This needs a newer version of raja, so you may have to rebuild some deps. I have found it useful to build with

--spec=" ^mfem~mpi~hypre"

Current In  == 0.7658220366
Current Out == 0.7658343481
Current In - Out == -1.231e-05

I.S. quadtree 13 levels, abs_tau_flux == 1.000e-04    size == 76376805

(apprx) Current In  == 0.7657741516
(apprx) Current Out == 0.7657625209
(apprx) Current In - Out == 1.163e-05
(store_samples needs to be fixed still)
Set use_eggs=false to try uniform absorption and see that it is preserved to machine precision on all domains.
While per-domain discrepancies are <1e-6, the total metric gives 1e-2.
This could indicate a problem with synchronizing shared boundary faces,
or it could be that my test is wrong.
Domain[59] excess== 0.00000000068
Domain[60] excess== 0.00000000306
Domain[61] excess==-0.00000000186
Domain[62] excess==-0.00000000186
Domain[63] excess== 0.00000000071
Entire     excess== 0.02622323856
  leak:1.68e+01  src:1.00e+00  rem:-1.58e+01
  cancelling==-0.02622259641
excess+cancelling==6.421469e-07

This shows that, for some reason, the sum over domains of sum_side_current()
of the problem-interior domain boundaries is not getting cancelled as it should.
…on and initialization.

Now the inflow boundaries have matching received and interpolated-from-received.
Conservation discrepancy down from 1e-2 to 6e-7.
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants