LeMat-Traj: A Scalable and Unified Dataset of Materials Trajectories for Atomistic Modeling

2025-08-28

AI Summary

Quantum-mechanical trajectory data—typically generated via density functional theory (DFT)—suffer from severe fragmentation in format, inconsistent metadata, and poor accessibility, hindering the development of high-accuracy machine-learned interatomic potentials (MLIPs). To address this, we introduce LeMat-Traj, the first large-scale, standardized trajectory dataset comprising over 120 million atomic configurations, spanning major DFT functionals and unifying data formats and metadata schemas. We propose a novel multi-source trajectory joint modeling framework and a synergistic filtering strategy for high-energy/high-force configurations, enabling unified training for both force-field fitting and structural relaxation pathways. Accompanying open-source tools—LeMaterial-Fetcher—support dynamic dataset expansion and community-driven curation. Fine-tuned MLIPs trained on LeMat-Traj achieve significantly reduced force prediction errors in structural relaxation tasks. Both LeMat-Traj and associated tooling are publicly released to foster open, reproducible, and collaborative atomic-scale modeling.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

The development of accurate machine learning interatomic potentials (MLIPs) is limited by the fragmented availability and inconsistent formatting of quantum mechanical trajectory datasets derived from Density Functional Theory (DFT). These datasets are expensive to generate yet difficult to combine due to variations in format, metadata, and accessibility. To address this, we introduce LeMat-Traj, a curated dataset comprising over 120 million atomic configurations aggregated from large-scale repositories, including the Materials Project, Alexandria, and OQMD. LeMat-Traj standardizes data representation, harmonizes results and filters for high-quality configurations across widely used DFT functionals (PBE, PBESol, SCAN, r2SCAN). It significantly lowers the barrier for training transferrable and accurate MLIPs. LeMat-Traj spans both relaxed low-energy states and high-energy, high-force structures, complementing molecular dynamics and active learning datasets. By fine-tuning models pre-trained on high-force data with LeMat-Traj, we achieve a significant reduction in force prediction errors on relaxation tasks. We also present LeMaterial-Fetcher, a modular and extensible open-source library developed for this work, designed to provide a reproducible framework for the community to easily incorporate new data sources and ensure the continued evolution of large-scale materials datasets. LeMat-Traj and LeMaterial-Fetcher are publicly available at https://huggingface.co/datasets/LeMaterial/LeMat-Traj and https://github.com/LeMaterial/lematerial-fetcher.

Problem

• Standardizing fragmented quantum mechanical trajectory datasets for materials
• Lowering barriers for training transferable machine learning interatomic potentials
• Providing harmonized atomic configurations across multiple DFT functionals

Innovation

• Standardizes data representation for atomic configurations
• Aggregates over 120 million configurations from multiple repositories
• Provides modular open-source library for reproducible data integration