Hi Peter In the 6.2 user manual (Section 4.6.3), there is a list of alternate implicit solvent methods supported by OpenMM. It says "OpenMM supports all of the Generalized Born models used by AMBER". Are we talking about the force field here or the MD simulation package? I want to try out different ...

Makes sense, and it now works as expected. Thanks a ton!

Oh I certainly *am setting the mass after creating the Context. I see, I didn't realize that cannot be changed. My initialization routine sets all the forces and masses, and after than I set the restraints, so probably thats my mistake!

So it may be easier to debug using constraints. Suppose I get the masses of all three atoms to be zero, I would expect the simulations to do nothing. However I still get a trajectory with the atoms moving around. After each integration step I printed out the masses using for step in xrange(STEPS): a...

I'll try that. I also hacked the above code to constrain it completely by replacing # force.addParticle(i, [p[0], p[1], p[2]]) self.openmm_system.setParticleMass(i, 0.0) just to see whether the O will move, and the Oxygen still moved everywhere so clearly I have not set up the forces correctly.

Hi all I'm trying to test out my restraints implementation using the CustomExternalForce as documented in the Python API (Openmm-6.2, Python API). My code is def restrain_term(self, restrain_atoms): # restrain_atoms is a list of atom indices to restrain, like [0,1] rest_atoms = [int(a) for a in rest...

Ok thanks Peter. Also in the example above, types 2 and 4 are the same atom type (CT), but the charges are different (though the GB radius and scale factor is the same). What does this imply?

Hi guys Great to see that the GBSA implicit solvent parameters are being well documented in the upcoming 6.2 release. I may be jumping the gun, but I downloaded the latest source, and was trying to find the parameters for the Amber99 force field. In the file amber99_obc.xml. I see the following entr...

I thought about nvidia-smi myself, but is that not CUDA specific? I mean is there a guarantee that device 0 from nvidia-smi is the same as device 0 from the OpenCL API?

Thanks Peter, that answers all my questions. I still have a problem with switching to OpenCL, but I think that is beyond OpenMM's scope. Consider a node with 2 GPUs. I rely on my cluster management software to schedule at maximum 2 MD jobs on this node, using a resource pool of 2 GPU tokens (the tok...