Grand-canonical MD simulation with H4D method

H4D, Hybrid 4D NEMD/MC method in conjunction with LAMMPS

• Author : Jeongmin Kim, a former postdoc at PHENIX with Benjamin Rotenberg
• Contact : jeongmin0658 AT gmail DOT com

Currently, JK leads the Kim research group at KENTECH, Naju, Korea.

---

The sources and examples are included:

H4D : LAMMPS c++ and head files

ver0.0 : H4D Python scripts

examples : Example systems, including

• LJ_salt100 : Example of LJ electrolytes
  • gcmd : Grand-canonical MD simulation
    • both-exchange: Exchange of both salt and solvent
    • salt-exchange: Exchange of a salt pair
    • solvent-exchange: Exchange of a solvent particle
  • sampling : Simulation to compute chemical potential (no accept of trial moves)
    • insertion : Insertion trial moves
    • deletion : Deletion trial moves
• JC_SPCE_1m : Example of aqueous NaCl electrolytes
  • gcmd : Grand-canonical MD simulation
    • salt-exchange: Exchange of a salt pair
    • solvent-exchange: Exchange of a solvent particle

Note that each folder also contains its own readme file.

---

References

J. Kim, L. Belloni, B. Rotenberg, J. Chem. Phys. 159, 144802 (2023)

L. Belloni, J. Chem. Phys. 151, 021101 (2019) - original H4D method

• Useful link Hybrid Monte Carlo with LAMMPS by Jeremy C. Palmer et al.

---

LAMMPS installation with H4D

Our H4D method works well with LAMMPS of a version of 27 Oct 2021. You can download lammps-27Oct2021.tar.gz.

H4D folder includes all the cpp and head files for lammps installation

• *_neq.cpp
• *_neq.h

How to use:

1. Copy this folder to the src folder for LAMMPS installation
2. Type make yes-H4D, before you install LAMMPS
3. Then, install LAMMPS as a shared library and Python module.

• Installation option should be mode=shared. See Manual 3.4.3
• For example, make mode=shared machine
• Then, make install-python. See Manual 2.2.1

Don't forget that you need other regular LAMMPS packages such as "Kspace", "MISC", etc. See Manual 3.6

---

H4D run

Before running LAMMPS with H4D, you must define where the "ver0.0" folder is located. You may use setup.sh in the folder where you execute the simulation.

run_gcmc_sample.py contains the parameters that control the H4D.

Simply run:

python3 run_gcmc_sample.py

or in case you use multiple CPUs,

mpirun -np number_of_cpus python3 run_gcmc_sample.py You can see the example_job_submission.sh.

---

LAMMPS outputs

All the outputs can be moved to the "output" folder. Use copy_output.sh.

• logfile_hneqMDMC.*: LAMMPS output of each "n_MC" (0 - n_MC-1).
• restart_after_MC_*.data: LAMMPS restart file (frequency determined by "save_freq").
• restart_end_run.data: LAMMPS restart file at the end of simulation.
• mc_statistics.out: contains H4D results such as energy changes, acceptance, etc.

---

Notes

• The code is not tested for Python versions other than Python3

---

Dependencies

H4D requires the following packages to run:

• numpy
• scipy
• math
• os
• sys
• random
• ctypes