Pymbar 3 4

[mikemhenry]

Description

Provide a brief description of the PR's purpose here.

Todos

• Implement feature / fix bug
• Add tests
• Update documentation as needed
• Update changelog to summarize changes in behavior, enhancements, and bugfixes implemented in this PR

Status

• Ready to go

Changelog message

[mikemhenry]

Supersedes #630

[mikemhenry]

hmmmm, seeing the same errors for pymbar 3&4

======================================================================
ERROR: Test multistate sampler on a harmonic oscillator without unsampled endstates
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/runner/micromamba-root/envs/openmmtools-test/lib/python3.8/site-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
  File "/home/runner/work/openmmtools/openmmtools/openmmtools/tests/test_sampling.py", line 234, in test_without_unsampled_states
    self.run(include_unsampled_states=False)
  File "/home/runner/work/openmmtools/openmmtools/openmmtools/tests/test_sampling.py", line 223, in run
    raise Exception("Dimensionless free energy difference exceeds MAX_SIGMA of %.1f" % MAX_SIGMA)
Exception: Dimensionless free energy difference exceeds MAX_SIGMA of 6.0

[mikemhenry]

That test passes fine on main but on this branch it fails ~10% of the time...

[mikemhenry]

I yelled when I saw this: https://github.com/choderalab/openmmtools/blob/main/openmmtools/multistate/utils.py#L178-L183

I could not figure out how an import fixed things, well it "solved" the problem that this bare except hid.

BUT what is crazier is that we don't get a failure every time, but will essentially calculate a subsample rate larger than than our input array, which causes our equilibrated indices to be indices=[0] which sometimes passes our tests if we are lucky...

Anyway, bare excepts were a bad idea. I can't belive the amount of time I spent on this.

[mikemhenry]

I will be cleaning up those exceptions in this PR as well....

[ijpulidos]

Oh yeah, of course bare excepts are biting us, good catch! Thanks.

[mikemhenry]

On windows-latest, py-3.8.10, OpenMM-8.0, pymbar-4

======================================================================
ERROR: Test expected number of entries in real time analysis output yaml file.
----------------------------------------------------------------------
Traceback (most recent call last):
  File "C:\Users\runneradmin\micromamba-root\envs\openmmtools-test\lib\site-packages\nose\case.py", line 197, in runTest
    self.test(*self.arg)
  File "D:\a\openmmtools\openmmtools\openmmtools\tests\test_sampling.py", line 1577, in test_real_time_analysis_yaml
    sampler.run()
  File "D:\a\openmmtools\openmmtools\openmmtools\multistate\multistatesampler.py", line 791, in run
    self._update_timing(iteration_time, partial_total_time, run_initial_iteration, iteration_limit)
  File "D:\a\openmmtools\openmmtools\openmmtools\multistate\multistatesampler.py", line 1773, in _update_timing
    self._timing_data["ns_per_day"] = iteration_simulated_nanoseconds / (
ZeroDivisionError: float division by zero

and on windows-latest, py-3.10, OpenMM-8.0, pymbar-3

======================================================================
ERROR: Test that creating a MultiState by storage string and reporter is the same
----------------------------------------------------------------------
Traceback (most recent call last):
  File "C:\Users\runneradmin\micromamba-root\envs\openmmtools-test\lib\site-packages\nose\case.py", line 197, in runTest
    self.test(*self.arg)
  File "D:\a\openmmtools\openmmtools\openmmtools\tests\test_sampling.py", line 1436, in test_storage_reporter_and_string
    sampler.run()
  File "D:\a\openmmtools\openmmtools\openmmtools\multistate\multistatesampler.py", line 791, in run
    self._update_timing(iteration_time, partial_total_time, run_initial_iteration, iteration_limit)
  File "D:\a\openmmtools\openmmtools\openmmtools\multistate\multistatesampler.py", line 1773, in _update_timing
    self._timing_data["ns_per_day"] = iteration_simulated_nanoseconds / (
ZeroDivisionError: float division by zero

All the other windows builds are fine, but we probably do want to catch that error so it doesn't crash a simulation if it happens. That can be done in another PR.

[mikemhenry]

Okay! Since everything is passing except those two errors, I am now going to switch up the CI matrix so that we are testing mostly with pymbar 4, but will keep a few jobs at pymbar 3 to make sure we maintain support for both.

[mikemhenry]

Once we finalize the matrix, I will mark them as required

[ijpulidos]

This is looking great. Just a few comments I think we need to adress.

[ijpulidos]

Do we want to raise the exception here? I feel like this might be disruptive for current users/applications.

[mikemhenry]

See #659 (comment) I think we can see if perses has any problems as well as other downstream packages as a double check. I worry it does more damage leaving it in the codebase. I could be convinced to just log an error but I think no one will see it.

[ijpulidos]

Looks great, thanks! I tested with perses locally in a tyk2 benchmark system and it seems to be working (using pymbar3). LGTM!