MLIPOps is a new project providing fast, portable implementations of common operations in machine learning potentials. Neighbor lists, Coulomb, dispersion, batch processing, and more. The first release is out and we’d love feedback! github.com/openmm/openm...

Introducing MLIPOps · openmm openmm · Discussion #5365

MLIPOps is a new package we've created. It provides PyTorch modules for lots of operations commonly used in machine learning interatomic potentials (MLIPs). The first public release is now out, and...

github.com

The beta of OpenMM 8.6 is now available for testing! This release has new methods for multistate sampling (replica exchange and expanded ensemble sampling), new barostat functionality, and performance improvements. Please test it out and let us know if you find any issues: github.com/openmm/openm...

OpenMM 8.6.0 beta · openmm openmm · Discussion #5345

A beta of OpenMM 8.6 is now available for testing. The most important new feature in this release is multistate sampling. ReplicaExchangeSampler and ExpandedEnsembleSampler implement algorithms for...

github.com

OpenMM 8.5.2 (github.com/openmm/openm...) and OpenMM-ML 1.7 (github.com/openmm/openm...) have been released. These releases include various bugfixes, including to an issue preventing PyTorch-based ML models from running in certain circumstances. PyPI and conda-forge packages are now available.

Release OpenMM 8.5.2 · openmm/openmm

This is a patch release containing the following fix. #5279: Make sure contexts are deselected before evaluation

github.com

Our paper on benchmarking pretrained machine learning potentials has been published in the Journal of Chemical Theory and Computation, and now includes updated performance data as well as more details on model accuracy. Read it at doi.org/10.1021/acs....

Accuracy and Efficiency Benchmarks of Pretrained Machine Learning Potentials for Molecular Simulations

The rapid development of pretrained Machine Learning Interatomic Potentials (MLIPs) that cover a wide range of molecular species has made it challenging to select the best model for a given applicatio...

doi.org

OpenMM 8.5 includes a new local energy minimizer implementation that runs entirely on the GPU and is used automatically on any of OpenMM’s GPU-accelerated platforms. Our benchmarks demonstrate a 4-6x speedup on systems of various sizes.

The LCPO (linear combination of pairwise overlaps) surface area method for implicit solvent simulations has been added to OpenMM 8.5. Add it to your simulation using ForceField.createSystem(sasaMethod='LCPO'). Read more in the user guide: docs.openmm.org/latest/userg...

OpenMM 8.5.1 has been released. This is a patch release containing a few bug fixes. Packages have been published to conda-forge and PyPI as usual; see also github.com/openmm/openm...

OpenMM-ML 1.6 has been released! It supports a much wider range of models (including MACE-OMOL-0, FeNNix-Bio1, and Orb-v3). You can also use OpenMM with any package that supports the ASE (Atomic Simulation Environment) Calculator interface. "pip install openmmml". github.com/openmm/openm...

GitHub - openmm/openmm-ml: High level API for using machine learning models in OpenMM simulations

High level API for using machine learning models in OpenMM simulations - openmm/openmm-ml

github.com

OpenMM 8.5.0 has been released! New features include GPU-accelerated energy minimization, support for custom Python computation of energies and forces, the LCPO surface area for implicit solvation, and more. Install "openmm" from PyPI or conda-forge, or get the source at github.com/openmm/openm...

Release OpenMM 8.5.0 · openmm/openmm

This release adds the following new features. The energy minimizer has been rewritten to run entirely on the GPU. This can make energy minimization dramatically faster. PythonForce is a new class...

github.com

The release candidate of OpenMM 8.5 is now available. If you try it out, please let us know of any issues you find; if no further problems are found, the release should follow in about a week. github.com/openmm/openm...

Planning for 8.5 release · openmm openmm · Discussion #5194

We're hoping to create the beta of 8.5 fairly soon, which means the window for getting new features into it is starting to close. Here are the new features that either have already been implemented...

github.com

If you're interested in trying out the beta of OpenMM 8.5 and haven't had a chance to do so yet, we're just over halfway through the beta period. Let us know of any issues you find so we can try to get them fixed before the release candidate comes out late next week. github.com/openmm/openm...

openmm/openmm

OpenMM is a toolkit for molecular simulation using high performance GPU code. - openmm/openmm

github.com

The beta of OpenMM 8.5 is now available! New features include energy minimization entirely on the GPU, an API that lets custom Python code compute forces, the LCPO surface area method for implicit solvation, and more. Please try it out and tell us if you find any problems. github.com/openmm/openm...

Planning for 8.5 release · openmm openmm · Discussion #5194

We're hoping to create the beta of 8.5 fairly soon, which means the window for getting new features into it is starting to close. Here are the new features that either have already been implemented...

github.com

We're planning for the next (8.5) version of OpenMM and hoping to make a beta release soon. Feel free to review and discuss the proposed list of new features at github.com/openmm/openm...

Planning for 8.5 release · openmm openmm · Discussion #5194

We're hoping to create the beta of 8.5 fairly soon, which means the window for getting new features into it is starting to close. Here are the new features that either have already been implemented...

github.com

Our new paper provides benchmarks of various pretrained machine learning interatomic potentials for accuracy, computational efficiency, memory use, and simulation stability. Read the preprint on ArXiv: arxiv.org/abs/2601.16331

Accuracy and Efficiency Benchmarks of Pretrained Machine Learning Potentials for Molecular Simulations

The rapid development of pretrained Machine Learning Interatomic Potentials (MLIPs) that cover a wide range of molecular species has made it challenging to select the best model for a given applicatio...

arxiv.org

Machine Learning Potentials are a powerful tool for high accuracy simulation. See our new tutorial on using them in OpenMM. openmm.github.io/openmm-cookb...

The constant potential method, a new feature in OpenMM 8.4, allows you to simulate electrodes and other conductive materials. To learn how it works and how to use it in OpenMM, we’ve added a tutorial at openmm.github.io/openmm-cookb....

OpenMM 8.4 now includes an implementation of the adaptive quantum thermal bath algorithm. To learn how to use it in your simulations, try out our new tutorial on nuclear quantum effects at openmm.github.io/openmm-cookb....

We’ve recently updated and refreshed the OpenMM Tutorials, which now include a series of four introductory tutorials designed to teach the basics of OpenMM to first-time users. Check them out at openmm.github.io/openmm-cookb....

OpenMM 8.4 has been released! New features in this version include adaptive quantum thermal bath, constant potential, Tinker file support, orientation restraints, radius of gyration calculation, and more. Install the "openmm" package from conda-forge and PyPI or get it at github.com/openmm/openm....

Release OpenMM 8.4.0 · openmm/openmm

This release adds lots of new features. QTBIntegrator implements the adaptive quantum thermal bath algorithm. This is a fast method of adding nuclear quantum effects to a simulation. ConstantPoten...

github.com

The release candidate of OpenMM 8.4 is now available. Please test it and see if it works. Assuming no further problems are found, the release will follow in about a week. Thanks to all who tried out the beta version to look for issues! github.com/openmm/openm...

OpenMM 8.4 beta · openmm openmm · Discussion #5087

The beta of OpenMM 8.4 is now available. You can install it with mamba/conda with the command mamba install -c conda-forge/label/openmm_rc -c conda-forge openmm or using pip with the command pip in...

github.com

The beta of OpenMM 8.4 is now available! Features coming to this release include adaptive quantum thermal bath, constant potential, Tinker input file format support, and more. Please test it out, try the new features, and tell us if you find any problems. github.com/openmm/openm...

OpenMM 8.4 beta · openmm openmm · Discussion #5087

The beta of OpenMM 8.4 is now available. You can install it with mamba/conda with the command mamba install -c conda-forge/label/openmm_rc -c conda-forge openmm or using pip with the command pip in...

github.com

We've started planning for a future 8.4 release of OpenMM! Feel free to visit our discussion forum to comment on proposed features and make suggestions: see github.com/openmm/openm....

Planning for 8.4 release · openmm openmm · Discussion #5051

It's time to start thinking about the 8.4 release. We should be ready to create the beta quite soon. A number of new features have already been added. We need to decide what remaining features to t...

github.com

OpenMM 8.3.1, a patch release containing fixes for a few issues present in OpenMM 8.3.0, has been released. It is available as "openmm" on conda-forge and PyPI, as well as at github.com/openmm/openm....

Release OpenMM 8.3.1 · openmm/openmm

This is a patch release containing the following fixes. #4980: Fixed bug in computing pressure #4986: Add function parsing to custom multiparticle force #4989: Update CHARMM36 2024 and add water mo...

github.com

We have created a release candidate for OpenMM 8.3.1, a patch release containing resolutions for a few issues present in the recent release of OpenMM 8.3.0. Please test it and tell us if you find any problems. github.com/openmm/openm...

Create 8.3.1 · Issue #4999 · openmm/openmm

A few bugs have come up in 8.3 that call for a patch release. It should definitely include #4980: Fixed bug in computing pressure #4989: Update CHARMM36 2024 and add water models I suggest also inc...

github.com

This blog post from NVIDIA describes how to use MPS with OpenMM to improve your overall simulation throughput. developer.nvidia.com/blog/maximiz...

Maximizing OpenMM Molecular Dynamics Throughput with NVIDIA Multi-Process Service | NVIDIA Technical Blog

Molecular dynamics (MD) simulations model atomic interactions over time and require significant computational power. However, many simulations have small (<400 K atoms) system sizes that underutilize…

developer.nvidia.com

OpenMM 8.3, including updated force fields, a new DPD (dissipative particle dynamics) integrator, pressure calculation, and other new features and improvements, has been released. It is available as "openmm" on conda-forge and PyPI, as well as at github.com/openmm/openm....

Release OpenMM 8.3.0 · openmm/openmm

This release adds a number of new features. Here are the more important ones. It adds several new or updated force fields. That includes Amber ff19SB, the 2024 version of CHARMM36, and the 2023 ve...

github.com

The release candidate of OpenMM 8.3 is now available. Please test it and see if it works. Assuming no problems are found, the release will follow in about a week. github.com/openmm/openm...

OpenMM 8.3 beta · openmm openmm · Discussion #4926

A beta of OpenMM 8.3 is now available. Please try it out and let us know how it works for you! You can install it with mamba/conda with the command mamba install -c conda-forge/label/openmm_rc -c c...

github.com

The beta of OpenMM 8.3 is now available! Please test it, try the new features, and tell us if you find any problems. github.com/openmm/openm...

OpenMM 8.3 beta · openmm openmm · Discussion #4926

A beta of OpenMM 8.3 is now available. Please try it out and let us know how it works for you! You can install it with mamba/conda with the command mamba install -c conda-forge/label/openmm_rc -c c...

github.com

OpenMM 8.3 will add support for Dissipative Particle Dynamics (DPD). Try it in the latest development build with "mamba install -c conda-forge/label/openmm_dev openmm". Documentation is online at docs.openmm.org/development/....

Benchmarks of OpenMM running on the NVIDIA RTX 5090 GPU are now available: for more information, visit the OpenMM benchmarks page at openmm.org/benchmarks.