You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We want to implement tests of ForceBalance's optimization process using liquid parameters. However, liquid simulations are not deterministic (even given same starting parameters, hardware differences may cause different results), and are prohibitively time-consuming.
When the property calculator API is finalized, we should store a small set of test trajectories, mirroring the simulations that would be run by ForceBalance in a specific optimization run. These simulation trajectories would be returned by a mock property-calculator interface. This setup would allow robust integration testing of ForceBalance's optimization using trajectory data.
The text was updated successfully, but these errors were encountered:
Per discussion at https://openforcefieldgroup.slack.com/archives/C8P8JEE5T/p1545071217055600
We want to implement tests of ForceBalance's optimization process using liquid parameters. However, liquid simulations are not deterministic (even given same starting parameters, hardware differences may cause different results), and are prohibitively time-consuming.
When the property calculator API is finalized, we should store a small set of test trajectories, mirroring the simulations that would be run by ForceBalance in a specific optimization run. These simulation trajectories would be returned by a mock property-calculator interface. This setup would allow robust integration testing of ForceBalance's optimization using trajectory data.
The text was updated successfully, but these errors were encountered: