Tutorials for the SOMD code

This repo contains some tutorials of the SOMD code. Going through these tutorials will make you familiar with SOMD quickly.

Contents

Basic tutorial

In this tutorial we will calculate the oxygen−oxygen radial distribution function of bulk waters. By finishing this tutorial, you may earn some basic sense about performing MD simulations with SOMD.

Building Neuroevolution Potentials with active learning, part I

In this tutorial we will try to fit non-reactive Neuroevolution Potentials (NEPs) of an isolated ethane molecule, by the mean of the active learning methodology. The metadynamics method will also be invoked to accelerate the configuration space sampling.

Building Neuroevolution Potentials with active learning, part II

In this tutorial we will try to fit reactive Neuroevolution Potentials (NEPs) for the symmetric $CH_3Cl + Cl^- \rightarrow Cl^- + CH_3Cl$ reaction within an explicit water box, by the mean of the active learning methodology. Steered MD and metadynamics methods will be invoked to accelerate the configuration space sampling.

How to run

First, make sure that each pre-required package has been installed. Then clone this repo:

git clone https://www.github.com/initqp/somd_tutorials

You can find instructions for each tutorial from the README.md file in the corresponding directory.

Pre-requirements

Before running these tutorials, make sure that following packages has been properly installed.

SOMD

Make sure that SOMD has been properly installed.

>>> import somd
>>> print(somd.__version__)
# version string here

If the package has been installed, a version string should be printed. Otherwise, follow the instructions in the main repo to install it.

DFTD3

Start a python REPL and enter following lines:

>>> import dftd3
>>> print(dftd3.__version__)
# 0.6.0

If the package has been installed, a version string should be printed. Otherwise, install it with:

conda install dftd3-python -c conda-forge

or

pip install dftd3

PLUMED

The enhanced sampling methods in SOMD are provided by the PLUMED package. Thus, to complete this tutorial, the PLUMED package and its python wrapper must be installed. Start a python REPL and enter following lines:

>>> import plumed
>>> plumed.Plumed()
# +++ Loading the PLUMED kernel runtime +++
# +++ PLUMED_KERNEL="/path/to/libplumedKernel.so" +++

If the packages have been properly installed, the path to the PLUMED kernel library will be printed. Otherwise, you should compile the PLUMED main package (following instructions here) or install it through Conda:

conda install plumed -c conda-forge

Once you have the PLUMED main package installed, you should set the PLUMED_KERNEL environment variable to the actual path of the libplumedKernel.so file:

export PLUMED_KERNEL=/path/to/libplumedKernel.so

You also may need to add the above command to your bashrc file for later usages (if you installed your PLUMED main package via Conda, this kernel library should be located at the library directory of Conda, e.g., /conda/prefix/lib/libplumedKernel.so). Finally, install the PLUMED python wrapper with following commands (if you could figure out how to compile the wrapper by yourself, just skip this step):

conda install py-plumed -c conda-forge

or

pip install plumed