Cross-functional transferability in universal machine learning interatomic potentials

2025-04-07

AI Summary

This work addresses the performance degradation of universal machine-learned interatomic potentials (uMLIPs) when transferring across density functionals (e.g., GGA → r²SCAN), caused by energy-scale misalignment and weak correlation between low- and high-fidelity data. We propose a novel multi-fidelity transfer learning paradigm that— for the first time—identifies the critical role of elemental energy reference frames in cross-functional transfer, introduces an element-wise energy calibration mechanism, and integrates multi-stage fine-tuning with scaling-law analysis. Evaluated on the MP-r²SCAN dataset (240K structures), our method significantly improves data efficiency at sub-million-structure scales, drastically reducing the high-fidelity data requirements for constructing accurate potential energy surfaces compared to ab initio training. The approach establishes a generalizable, reproducible pathway for cross-functional knowledge reuse in uMLIPs, enabling robust transfer from widely available GGA-level data to higher-accuracy r²SCAN-level predictions.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Multiagent Systems: Multiagent Learning
• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Machine learning and data science for the Web

Abstract

The rapid development of universal machine learning interatomic potentials (uMLIPs) has demonstrated the possibility for generalizable learning of the universal potential energy surface. In principle, the accuracy of uMLIPs can be further improved by bridging the model from lower-fidelity datasets to high-fidelity ones. In this work, we analyze the challenge of this transfer learning problem within the CHGNet framework. We show that significant energy scale shifts and poor correlations between GGA and r$^2$SCAN pose challenges to cross-functional data transferability in uMLIPs. By benchmarking different transfer learning approaches on the MP-r$^2$SCAN dataset of 0.24 million structures, we demonstrate the importance of elemental energy referencing in the transfer learning of uMLIPs. By comparing the scaling law with and without the pre-training on a low-fidelity dataset, we show that significant data efficiency can still be achieved through transfer learning, even with a target dataset of sub-million structures. We highlight the importance of proper transfer learning and multi-fidelity learning in creating next-generation uMLIPs on high-fidelity data.

Problem

• Challenges in cross-functional data transferability in universal ML interatomic potentials
• Energy scale shifts and poor correlations between GGA and r2SCAN
• Importance of elemental energy referencing in transfer learning

Innovation

• Uses CHGNet framework for transfer learning
• Implements elemental energy referencing technique
• Combines multi-fidelity datasets for efficiency