Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add small molecule repex consistency tests (#1065)
* add neutral test * add small molecule tests * remove charge test * add charge test * add guest mol2s * fix charge changing test * add cache * make sure lrc is set correctly * Small code cleanup. Fix docstring. Remove unused imports. * Using LangevinDynamicsMove instead of LangevinSplittingDynamicsMove. Adds gpu tests decorator. * Only using LangevinDynamicsMove in tests. Others stay the same as before. * Making sure the CPU platform is used in geometry engine. * Adjusting repex tests parameters for better performance. * Revert "Making sure the CPU platform is used in geometry engine." This reverts commit 1e90291. * Keep the convergence criteria to be 6*dDDG * Increasing n steps for move apply. Decreasing number of iterations * Trying running self-hosted GPU tests on PR. * switch to using openmm rc * Testing phenol-paracetamol transformation. 250 steps, 1000 iterations. Oh Percy... y u no like smiles! * Quick testing new instance type for gpu tests. * Trying own fork with fix with unique instance name * Going back to original runner repo. * add us-east-1a subnet * fix vim typo * try p2.xlarge first * Added us-east-1b subnet * Added us-east-1c subnet * Added us-east-1e subnet * Added us-east-1f subnet * go to next step even if we can't provision an instance * switch to g4dn.xlarge * switch to using once subnet for now while testing new ami * fix comment error * make sure we stop instances if they fail * bump ci * bump ci * bump ci * switch to using new ami * try new subnet and sg * bump ci * bump ci * switch to using micromamba * remove unused channels and clean up CI workflow a bit * debug provision-with-micromamba * see if python 3.10 fixes it * try latest version * try older version * Set default shell * switch to newest version * switch to a different action * See if this fixes the unbound home var * see if this fixes the issue with home not being set * Testing GPU platforms specifically, debugging purposes * timeout of 20 hrs instead of the 6 hrs default * back to testing everything * Putting timeout of 20 hrs in correct place/job. * Using specific tagged url for tyk2 tests * fix cudatoolkit version * remove extra code we don't need anymore * skip test for now, need better system to test with * let codecov fail, also update the action * only run on commit, on demand, on weekly * Using bootstrapping from MBAR to compute errors * Those damn signs! Making it less prone to sign errors. * Better naming for quantities * Making sure we test using decorrelated samples * Swapping tests, now correct order. * use micromamba * micromamba thowing a fit about oPtIonS iT dOeSn'T uNDerStAnd * Changing back the transformation order to the correct one. * Correct comparison charge mutation. Naming cleanup * Better naming. Minor indenting cleanup. * Removing misplaced quotes, my bad! * Update .github/workflows/self-hosted-gpu-test.yml * Removing deprecated comment * Lowering the number of steps per iteration to decrease run times. * removing commented/unused line * Update .github/workflows/self-hosted-gpu-test.yml --------- Co-authored-by: Iván Pulido <[email protected]> Co-authored-by: Iván Pulido <[email protected]> Co-authored-by: Mike Henry <[email protected]>
- Loading branch information