Selim Sami

I'm not sure if it's a good idea to increase to convergence criteria so much btw. In DrudeSCF you're under the assumption that forces on drude particles are (almost) zero...

As far as I know the forces should be a lot smaller than that. But I might be remembering something wrong. I'm quite puzzled why openmm and gromacs force convergence...

Thanks for the explanation, that's very illuminating. > If you want to do geometry optimization, use the energy minimizer. If you want to optimize just the Drude particles while keeping...

Aha. When I set Drude mass to 0 and then do: `simulation.context.setVelocitiesToTemperature(300*kelvin)` all velocities, including atoms, are zero. But then: `system.addForce(AndersenThermostat(300*kelvin, 1/picosecond))` seems to work correctly - it gradually increases...

Thanks Peter, 2nd option is also what I thought about but I was hoping to handle things on the FF side rather than the System side. Regarding the 1st option:...

Thanks @mattwthompson, that's very insightful. Just to clarify, the website you shared says that this code is not production ready. so is the ff14SB SMIRNOFF port ready to be used...

Hi Peter, I was referring to the later AMOEBA+ models where they started introducing the charge flux: https://pubs.acs.org/doi/full/10.1021/acs.jpclett.9b03489 https://pubs.aip.org/aip/jcp/article/153/6/064103/199466/Accurate-description-of-molecular-dipole-surface