Make MELD forces work with PBC

[jlmaccal]

I'm not sure what the current status of MELD forces and PBC is. I suspect that some cases may work fine, while others cause serious problems. OpenMM isn't super consistent about how PBCs are handled in forces, and MELD was never designed with PBC in mind.

The first step is to create a series of test cases:

• all of the different types of forces
• cases where the forces are within a single molecule
• cases where the forces are between molecules

Once we have some tests, we'll know what needs to be fixed.

[jlmaccal]

First steps:

• [ ] Add tests for all forces calculated on the GPU for both intra and intermolecular restraints with PBC.
• [ ] Add tests for all forces that can be calculated on the CPU for both intra and intermolecular restraints with PBC.

The expectation is that all of these tests should fail initially. We can then go about fixing the code to make the tests pass.

[frank-stonybrook]

Hi Justin, Thanks for your response. I will take a look at all the other forces in Meld and try to make some test cases. The test for intermolecular distance restraint is already included into TestMeldCudaForce.cpp in the PR. Best, Cong