Learning Memory and Transferability in Coarse-Grained Dynamics

This repository contains the complete dataset and source code supporting the findings of the manuscript “Learning Memory and Transferability in Coarse-Grained Dynamics”. We introduce a Transferable Non-Markovian Intelligent Dissipative Particle Dynamics (TNM-IDPD) framework, which solves the dual challenge of recovering accurate dynamics and achieving model transferability across thermodynamic states in coarse-grained molecular simulations. The dataset includes: Deep Neural Network (DNN) and Bayesian Neural Network (BNN) training codes and results for learning the force field and memory kernel. Input files and results for all-atom Molecular Dynamics (MD) and Non-Markovian DPD (NMDPD) simulations. Key results, validated on a star polymer system, demonstrate that our framework accurately captures both structural and dynamic properties. All files are organized in a clear directory structure documented in the included README.mdfile, which provides detailed file descriptions. How to use this dataset: Please refer to the README.mdfile in the root directory for comprehensive documentation. If this data is used, please cite both the associated paper and this dataset using the provided DOI.