Predicting Atomistic Transitions with Transformers

TL;DR

This paper demonstrates that transformer models can be trained to predict atomistic transition pathways in nano-clusters, offering a faster alternative to traditional simulation methods.

Contribution

The study introduces a transformer-based approach for predicting atomistic transitions, including methods for evaluating physical validity and generating diverse microstates.

Findings

Transformers can accurately predict atomistic transition pathways.

The model enables generation of multiple microstates through slight data variations.

Predictions are validated for physical plausibility.

Abstract

Accurate knowledge of the atomistic transition pathways in materials and material surfaces is crucial for many material science problems. However, conventional simulation techniques used to find these transitions are extremely computationally intensive. Even with large-scale, accelerated material simulations, the computational cost constrains the applicable domain in practice. Machine learning models, with the potential to learn the complex emergent behaviors governing atomistic transitions as a fast surrogate model, have great promise to predict transitions with a vastly reduced computational cost. Here, we demonstrate how transformers can be trained to predict atomistic transitions in nano-clusters. We show how we evaluate physical validity of the predictions and how a multitude of additional, different microstates can be generated by slightly varying the data provided to the model.