hymd 2.2.0

Massively parallel hybrid particle-field MD

---

HylleraasMD (HyMD) is a massively parallel Python package for Hamiltonian hybrid particle-field molecular dynamics (HhPF-MD) simulations of coarse-grained bio- and soft-matter systems.

HyMD can run canonical hPF-MD simulations, or filtered density Hamiltonian hPF (HhPF-MD) simulations [1],[2],[3] with or without explicit PME electrostatic interactions. It includes all standard intramolecular interactions, including stretching, bending, torsional, and combined bending-dihedral potentials. Additionally, topological reconstruction of permanent peptide chain backbone dipoles is possible for accurate recreation of protein conformational dynamics. It can run simulations in constant energy (NVE), constant volume (NVT) [1] or constant pressure (NPT) conditions [4].

HyMD uses the pmesh library for particle-mesh operations, with the PPFT [5] backend for FFTs through the pfft-python bindings. File IO is done via HDF5 formats to allow MPI parallel reads.

If you use HyMD, please cite our paper(s).

User Guide

Detailed installation and user guide, together with comprehensive example simulations are located in the HylleraasMD documentation.

Run simulations by

python3 -m hymd [CONFIGURATION_FILE] [TOPOLOGY_FILE] (--OPTIONAL_ARGS)

Run interactively in Google Colaboratory

A Google Colaboratory jupyter notebook is setup here with a working HyMD fully installed and executable in the browser.

Installation

Non-Python dependencies

HyMD installation requires a working MPI compiler. It is highly recommended to have MPI-enabled HDF5 and h5py for running parallel simulations with HyMD. Install both on Ubuntu with

sudo apt-get update -y
sudo apt-get install -y pkg-config libhdf5-mpi-dev libopenmpi-dev
python3 -m pip uninstall h5py  # Remove any serial h5py installation present
CC="mpicc" HDF5_MPI="ON" python3 -m pip install --no-binary=h5py h5py

Note There might be memory leaks if you use OpenMPI <= 4.1.1. See #186 for more details.

Python dependencies

Install HyMD with pip by

python3 -m pip install --upgrade numpy mpi4py cython
python3 -m pip install "pmesh @ git+https://github.com/rainwoodman/pmesh"
python3 -m pip install hymd

pmesh is installed from the GitHub repository because fixes to be compatible with modern NumPy versions were not pushed to PyPI. See HyMD docs for more information, including install steps for macOS and non-Debian linux distributions.

Run in docker

Alternatively, an up-to-date docker image is available from docker hub

docker pull mortele/hymd
docker run -it mortele/hymd
/app$ python3 -m pip install hymd
/app$
/app$ # Grab example input files
/app$ curl -O https://raw.githubusercontent.com/Cascella-Group-UiO/HyMD-tutorial/main/ideal_chain/ideal_chain.toml
/app$ curl -O https://raw.githubusercontent.com/Cascella-Group-UiO/HyMD-tutorial/main/ideal_chain/ideal_chain.HDF5
/app$
/app$ # Run simulation
/app$ python3 -m hymd ideal_chain.toml ideal_chain.HDF5 --verbose

Run tests

Clone the repository and run tests with pytest

git clone https://github.com/Cascella-Group-UiO/HyMD.git hymd
cd hymd
python3 -m pip install pytest pytest-mpi
pytest

Running MPI enabled pytest tests is simplified with a convenient script

chmod +x pytest-mpi
pytest-mpi -oo -n 2 -ns

Contributions and issues

We welcome contributions to our code and provide a set of guidelines to follow in CONTRIBUTING.md.
To seek support in case of any issues and bugs, we welcome you to post them using the issue tracker.

Please cite our work

You will find information about our publications and archived data since 2023 at the open repository: Publications.
If you use HyMD, please cite:

@article{
  Ledum_HylleraasMD_Massively_parallel_2023,
  author = {Ledum, Morten and Carrer, Manuel and Sen, Samiran and Li, Xinmeng and Cascella, Michele and Bore, Sigbjørn Løland},
  doi = {10.21105/joss.04149},
  journal = {Journal of Open Source Software},
  month = {apr},
  number = {84},
  pages = {4149},
  title = {{HylleraasMD: Massively parallel hybrid particle-field molecular dynamics in Python}},
  url = {https://joss.theoj.org/papers/10.21105/joss.04149},
  volume = {8},
  year = {2023}
}

@article{
   Sen_HylleraasMD_2023,
   author = {Sen, Samiran and Ledum, Morten and Bore, Sigbjørn Løland and Cascella, Michele},
   title = {Soft Matter under Pressure: Pushing Particle–Field Molecular Dynamics to the Isobaric Ensemble},
   doi = {10.1021/acs.jcim.3c00186},
   journal = {Journal of Chemical Information and Modeling},
   month= mar,
   year = {2023},
   volume = {63},
   number = {7},
   pages = {2207-2217},
   URL = {https://doi.org/10.1021/acs.jcim.3c00186},
}

---

References

[1] Ledum, M.; Sen, S.; Li, X.; Carrer, M.; Feng Y.; Cascella, M.; Bore, S. L. HylleraasMD: A Domain Decomposition-Based Hybrid Particle-Field Software for Multi-Scale Simulations of Soft Matter. ChemRxiv 2021

[2] Ledum, M.; Carrer, M.; Sen, S.; Li, X.; Cascella, M.; Bore, S. L. HyMD: Massively parallel hybrid particle-field molecular dynamics in Python. Journal of Open Source Software (JOSS) 2023, 8(84), 2475-9066, 4149.

[3] Bore, S. L.; Cascella, M. Hamiltonian and alias-free hybrid particle–field molecular dynamics. J. Chem. Phys. 2020, 153, 094106.

[4] Sen, S.; Ledum, M.; Bore, S. L.; Cascella, M. Soft Matter under Pressure: Pushing Particle–Field Molecular Dynamics to the Isobaric Ensemble. J Chem Inf Model 2023, 63(7), 1549-9596.

[5] Pippig, M. PFFT: An extension of FFTW to massively parallel architectures. SIAM J. Sci. Comput. 2013, 35, C213–C236.