SIMPLE to sample uniformly in s between sbeg(1) and sbeg(2) if num_surf=2#44
SIMPLE to sample uniformly in s between sbeg(1) and sbeg(2) if num_surf=2#44jlion-pf wants to merge 2 commits into
Conversation
|
Thanks @jlion-pf and sorry for the late response. I see the intent and would prefer to solve this in an even more general way as follows: Re-introduce a switch Really clean alternative: Introduce an abstract class or subroutine interface that allows multiple samplers, e.g. What do you think? Do you have capacity to do that? Or @dannythecore ? Otherwise it could take a while for myself. Best |
|
This works now with startmode=5 and setting two flux surfaces sbeg(1) and sbeg(2). |
## Summary Follow-up to #323. The squash-merged classification fix means main now matches the output that this branch's `simple.x` produces for `golden_record_classifier_fast`, so the temporary `GOLDEN_RECORD_SKIP_CASES=classifier_fast` escape hatch added in #323 can be removed. Two small changes: 1. `test/tests/CMakeLists.txt`: drop `GOLDEN_RECORD_SKIP_CASES=classifier_fast` from the per-test ENVIRONMENT. The `SKIP_CASES` plumbing in `compare_golden_results.sh` stays in place for future intentional-divergence bug fixes (just don't set the env var unless a fix needs it). 2. `test/golden_record/compare_golden_results.sh`: drop `avg_inverse_t_lost.dat` from the classifier file list. That file is only written when at least one particle is actually lost; with the fast_class fix the small `classifier_fast` test loses zero particles, so neither ref nor cur writes it. The comparator's "reference file missing" check then spuriously fails even though both runs agree byte-for-byte on the files that do exist. ## Verification Full regression suite, against freshly rebuilt main reference (cached `runs/run_main` and `simple_main` rebuilt to pick up #323): ``` $ make test-regression ... 13/13 Test #50: golden_record_sanity .................... Passed 0.57 sec 100% tests passed, 0 tests failed out of 13 Total Test time (real) = 411.09 sec ``` Both classifier cases: ``` 1/1 Test #45: golden_record_classifier_fast ... Passed 16.73 sec 1/1 Test #44: golden_record_classifier_combined ... Passed 16.18 sec ``` ## Test plan - [x] `make test-regression` passes 13/13 locally - [x] `golden_record_classifier_fast` (the test #323 had to skip) passes - [x] `golden_record_classifier_combined` (new in #323) still passes
2 participants
With this change, SIMPLE will start to uniformly sample between
sbeg(1)andsbeg(2)if the user providesnum_surf=2in the input file.Happy to discuss different implementation of this feature @krystophny