Learning Macroscopic Dynamics from Partial Microscopic Observations

Authors: Mengyi Chen, Qianxiao Li

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the accuracy, force computation efficiency, and robustness of our method on learning macroscopic closure models from a variety of microscopic systems, including those modeled by partial differential equations or molecular dynamics simulations.
Researcher Affiliation Academia Mengyi Chen1, Qianxiao Li1, 2 Department of Mathematics, National University of Singapore1, Institute for Functional Intelligent Materials, National University of Singapore2 chenmengyi@u.nus.edu, qianxiao@nus.edu.sg
Pseudocode Yes Algorithm 1 Data generation. and Algorithm 2 Training procedure.
Open Source Code Yes Our code is available at https://github.com/MLDS-NUS/Learn-Partial.git.
Open Datasets No The paper describes generating data from simulated systems (Predator-Prey, Lennard-Jones, Allen-Cahn) but does not provide access information or citations for a publicly available, pre-existing dataset. It states 'We choose D to be the trajectory distribution of the state x.'.
Dataset Splits No The paper discusses 'training data' and 'test dataset' but does not explicitly specify distinct 'validation' splits or percentages.
Hardware Specification Yes All the experiments are run on a single NVIDIA Ge Force RTX 3090 GPU.
Software Dependencies No The paper mentions LAMMPS for simulations but does not provide a specific version number. Other software dependencies like deep learning frameworks are implied but not explicitly versioned.
Experiment Setup Yes In our experiment, we choose a 3, b 0.4, λ 0. The microscopic equation is solved with a uniform time step t 0.01 from t 0 to t 30 using the Euler method. and λcond is a hyperparameter to adjust the ratio of Lcond and is chosen to be quite small in our experiments, e.g., 10 5 or 10 6. and The integration step is 0.001 and each trajectory is integrated for 250 steps. We sample the initial temperature randomly from r0.5, 1.5s.