ACEMD#
High-performance GPU molecular dynamics with a minimal configuration surface. Write a short YAML file, type acemd, and you’re running production MD on your GPU. The same configuration covers classical force fields (CHARMM, AMBER, OpenMM XML), hybrid neural-network potentials (NNP/MM), and pure NNP simulations.
ACEMD integrates with HTMD for system preparation and analysis from Python. Widely used in academic and industrial research — 700+ citations — and free for non-commercial use, with technical support for both academic and commercial users.
Step-by-step lessons. Start here if you’re new to ACEMD.
Task-focused recipes. “How do I X?”
CLI flags, input-file options, and the full Python API.
Concepts and design rationale — what ACEMD does and why.
Installation#
pip install "acemd[cu13]"
See Installation for CUDA-version matching, conda, and NNP-enabled variants.
Quick start#
Create an input.yaml describing the system and the run:
structure: dhfr.prmtop
coordinates: dhfr.pdb
boxsize: [62.23, 62.23, 62.23]
thermostat: true
run: 10ns
Then start the simulation:
acemd
ACEMD writes the trajectory to output.xtc, log values to output.csv, and a checkpoint to restart.chk. See tutorials/01-first-simulation for the full walkthrough.
Support#
Bug reports and questions go to GitHub issues for non-commercial users, or the Acellera helpdesk for commercial users. See Support for details.
Citing#
If ACEMD is useful in your research, please cite:
M. J. Harvey, G. Giupponi, and G. De Fabritiis. ACEMD: Accelerating biomolecular dynamics in the microsecond time scale. J. Chem. Theory Comput. 2009, 5 (6), 1632–1639. doi:10.1021/ct9000685