Closed mikemhenry closed 7 months ago
Thanks! No idea why the changes in #1242 didn't work, I'll be closing that.
With regards the GPU, I don't think the 8.1 rc is getting picked up by it, as per https://github.com/choderalab/perses/actions/runs/6894390986/job/18756206563#step:7:287
Haha that is my bad, I didn't even update the yaml file for the gpu jobs :facepalm:
Okay it is running on the GPU now https://github.com/choderalab/perses/actions/runs/6896218977/job/18761970375#step:7:314
(ignore the CPU tests right now, they passed already so its okay I made a mistake)
@ijpulidos okay looks like 2 errors on the GPU tests: https://github.com/choderalab/perses/actions/runs/6896894039/job/18764058862
One error looks like an API bug, the other though might be real:
FAILED perses/tests/test_repex.py::test_RESTCapableHybridTopologyFactory_repex_charge_mutation - pymbar.utils.ParameterError: Warning: Should have \sum_n W_nk = 1. Actual column sum for state 0 was 275.917719. 38 other columns have similar problems.
This generally indicates the free energies are not converged.
The tests work on main quite reliably, so we should investigate this IMHO, the first thing I will do is try and re-run the test to see if it happens again
re-running things here: https://github.com/choderalab/perses/actions/runs/6896894039
@mikemhenry it's actually something we have encountered before https://github.com/choderalab/pymbar/issues/419 I'm still not sure why is this happening, but I think it's only showing with pymbar4
We really need to track down that pymbar bug
GPU test here https://github.com/choderalab/perses/actions/runs/6894390986