Force-Aware Neural Tangent Kernels for Scalable and Robust Active Learning of MLIPs

TL;DR

This paper introduces a scalable, force-aware neural tangent kernel framework for active learning of MLIPs, enabling efficient screening of large candidate pools and robustness to biased distributions.

Contribution

It develops a novel force-aware NTK and a scalable acquisition framework that significantly improves active learning efficiency and robustness for MLIPs.

Findings

Achieves screening of ~200k structures within hours.

Force NTK yields lowest MAE and RMSE on OC20 dataset.

Remains robust under candidate-pool shift scenarios.

Abstract

Active learning for machine-learning interatomic potentials (MLIPs) must address several challenges to be practical: scaling to large candidate pools, leveraging energy-force supervision, and maintaining robustness when candidate pools are biased relative to the target distribution. In this work, we jointly address these challenges. We first introduce a linearly scaling acquisition framework based on chunked feature-space posterior-variance shortlisting. By avoiding materialisation of the candidate and train set kernels, this approach enables screening of ~200k structures within hours and applies broadly to acquisition strategies that score candidates based on molecular similarity metrics. We then extend the Neural Tangent Kernel (NTK) to a force-aware setting via mixed parameter-coordinate derivatives, yielding a force NTK and a joint energy-force NTK that provide natural similarity metrics for vector-field prediction. We demonstrate the effectiveness of the joint energy-force NTK on the OC20 dataset, where force-aware acquisition is crucial: it achieves the lowest energy and force MAE and RMSE across all metrics and distribution splits. Across T1x, PMechDB, and RGD benchmarks, our force NTK methods remain competitive with established baselines while being significantly more efficient than committee-based approaches. Under a controlled candidate-pool shift case study on T1x, acquisition based on pretrained MLIP embeddings and NTKs remains robust, whereas committee-based methods exhibit higher variance. Overall, these results show that a single pretrained MLIP can enable scalable, force-aware, and distribution-robust active learning for foundation-model fine-tuning.