Search found 2546 matches

by Peter Eastman
Mon Oct 09, 2023 4:08 pm
Forum: OpenMM
Topic: OpenMM 8.1 beta
Replies: 0
Views: 414

OpenMM 8.1 beta

The beta of OpenMM 8.1 is now available. You can install it with conda using the command conda install -c conda-forge/label/openmm_rc -c conda-forge openmm This release contains many performance improvements, particularly to the CUDA and OpenCL platforms. The largest speedups are for very large syst...
by Peter Eastman
Mon Oct 02, 2023 8:34 am
Forum: OpenMM
Topic: Python API to evaluate ewald Coulomb force
Replies: 14
Views: 1187

Re: Python API to evaluate ewald Coulomb force

If you want the particles to be completely immobile, you can just set their masses to zero. That will prevent them from moving at all.
by Peter Eastman
Thu Sep 28, 2023 12:31 pm
Forum: OpenMM
Topic: Benchmark of OpenMM on NVIDIA A10 cards
Replies: 1
Views: 266

Re: Benchmark of OpenMM on NVIDIA A10 cards

I don't know of any specific numbers for it, but I'd expect the A10 to be significantly slower than the others.
by Peter Eastman
Fri Sep 22, 2023 3:22 pm
Forum: OpenMM
Topic: What type of bond determines a molecule in a barostat?
Replies: 2
Views: 300

Re: What type of bond determines a molecule in a barostat?

The technical answer is that every Force class (or still more technically, the ForceImpl it creates) defines a getBondedParticles() method. It reports which particles are bonded to each other by that Force. In practice, the Forces that define bonds are HarmonicBondForce, CustomBondForce, CustomCentr...
by Peter Eastman
Thu Sep 21, 2023 11:58 am
Forum: OpenMM
Topic: Python API to evaluate ewald Coulomb force
Replies: 14
Views: 1187

Re: Python API to evaluate ewald Coulomb force

I modified your script to loop over all platforms and time the force evaluation for each one. import numpy as np import openmm as mm from openmm import app from openmm.unit import angstrom import time positions = np.random.rand(110, 3) * 100*angstrom # define the box size box_size = 100*angstrom # d...
by Peter Eastman
Thu Sep 21, 2023 11:06 am
Forum: OpenMM
Topic: Python API to evaluate ewald Coulomb force
Replies: 14
Views: 1187

Re: Python API to evaluate ewald Coulomb force

Another issue is your box size. 1000 nm is huge. The cost of PME scales with the size of the box. Do you really intend it to be that big?
by Peter Eastman
Thu Sep 21, 2023 7:59 am
Forum: OpenMM
Topic: Python API to evaluate ewald Coulomb force
Replies: 14
Views: 1187

Re: Python API to evaluate ewald Coulomb force

What are you timing? The whole script, or just the call to getState()? All the setup is much more expensive than the force evaluation. Creating a Simulation involves a lot of work, but it only has to be done once. The very first call to getState() often also has to do additional initialization. That...
by Peter Eastman
Mon Sep 11, 2023 8:31 am
Forum: OpenMM
Topic: Regarding addInteractionGroup() in CustomNonBodedForce
Replies: 1
Views: 259

Re: Regarding addInteractionGroup() in CustomNonBodedForce

Modifying a System or the Forces it contains has no effect on any Simulations that already exist. You need to make the changes before you create the Simulation. Alternatively you can call simulation.context.reinitialize().
by Peter Eastman
Fri Aug 25, 2023 8:44 am
Forum: OpenMM
Topic: Multiple CustomNonbondedForce
Replies: 1
Views: 314

Re: Multiple CustomNonbondedForce

This error is unrelated to having multiple CustomNonbondedForces. It means a single CustomNonbondedForce has had multiple exclusions added to it for the same pair of particles.
by Peter Eastman
Fri Aug 18, 2023 2:54 pm
Forum: OpenMM
Topic: Potential Energy error at Folding@Home
Replies: 1
Views: 331

Re: Potential Energy error at Folding@Home

You need to ask the Folding@home developers about this. I don't think they read this forum.