Research data supporting: "Relevant, hidden, and frustrated information in high-dimensional analyses of complex dynamical systems with internal noise"

This repository contains the set of data shown in the paper "Relevant, hidden, and frustrated information in high-dimensional analyses of complex dynamical systems with internal noise", published on arXiv (DOI: 10.48550/arXiv.2412.09412).

The scripts contained herein are:

 PCA-Analysis.py:  python script to calculate the SOAP descriptor, denoising it, and compute the Principal Component Analysis

SOAP-Component-Analysis.py: python script to calculate the variance of the single SOAP components

Hierarchical-Clustering.py: python script to compute the hierarchical clustering and plot the dataset

OnionClustering-1d.py: script to compute the Onion clustering on a single SOAP component or principal component

OnionClustering-2d.py: script to compute bi-dimensional Onion clustering

OnionClustering-plot.py: script to plot the Onion plot, removing clusters with population <1%

UMAP.py: script to compute the UMAP dimensionality reduction technique

Additional data contained herein are:

starting-configuration.gro: gromacs file with the initial configuration of the ice-water system

traj-ice-water-50ns-sampl4ps.xtc: trajectory of the ice-water system sampled every 4 ps

traj-ice-water-50ns-sampl40ps.xtc: trajectory of the ice-water system sampled every 40 ps

some files containing the SOAP descriptor of the ice-water system: ice-water-50ns-sampl40ps.hdf5, ice-water-50ns-sampl40ps_soap.hdf5, ice-water-50ns-sampl40ps_soap.npy, ice-water-50ns-sampl40ps_soap-spavg.npy

PCA-results: folder that contains some example results of the PCA

UMAP-results: folder that contains some example results of UMAP

The data related to the Quincke rollers can be found here: https://zenodo.org/records/10638736