Multiple timestep integrator and nonbondedCutoff

[Aron Broom]

Hi,

I'm wondering if it is possible to use the Multiple timestep integrator to evaluate forces outside the nonbondedCutoff less frequently than the other forces?

Since the identity of forces outside this cutoff would change over time, it seems the forcegroups would have to be updated every X timesteps or something. Maybe this wouldn't be worth it computationally?

Thanks,

Aron

[Peter Eastman]

You can create two different NonbondedForces, one with a short cutoff and one with a longer cutoff. The inner timesteps would use Fshort, and the outer timesteps would use Flong-Fshort. MTSIntegrator doesn't support that directly, since there's no way to tell it to use the difference between two forces. But it's just a CustomIntegrator, so you can easily create your own version that does that.

Peter