The code supplied here can be used to classify and characterize solvation einvironments in the condensed phase, based on all-atom Lammps simulations. Details of this work are published at zzzzz. The analysis for ProDOT-2Hex neat, presented in the paper, is uploaded here in the example directory.
The project consists of three components:
-
generating single-ion trajectories, starting from the an equilibrium simulations using genMDinput.m
The code takes as input seed.data and Li.data from example/input, which are Lammps topology files.
The code generates md-*.data files where the ion (in this case Li) is initialized at random wihtin the equilibrium seed simulation box.
The trajectories are carried out with Lammps, using the template input script supplied in lammps-scripts. Each run generates *.rdf *.eng and *.msd files which correspond to radial distribution functions, binding energy and mean square displacement.
A simple vmd script is supplied (vmd-scripts) to visualize the Lammps data files using: vmd -e visdata.tcl -args illus.data. -
reading the output trajectories and packinging the data into objects vecs using anlMDoutput.m
The code reads the Lammps output from example/MD-sims and creates object vecs which is written to example/SEC-ML -
classifying the solvation environments (SEs) based on rdf feature vectors using classifySEs.py
The code reads the vecs object and uses UMAP to embed the data in a 2D latent space.
The code then labels the clusters and obtains a classification of SEs.
The code clusters the rdfs, msds, engs based on SE membership and creates useful plots in example/SEC-ML