aiqbench analysis
ADAPT introduces a novel machine learning force field that replaces graph neural networks with a direct coordinates-in-space formulation to better model long-range interactions in materials. By treating atoms as tokens within a Transformer architecture, this approach significantly reduces computational costs while improving prediction accuracy for point defects compared to existing GNN-based models. This advancement addresses critical limitations in current MLFFs, offering a more efficient pathway for high-throughput defect database analysis and structural relaxation simulations.
ADAPT: Lightweight, Long-Range Machine Learning Force Fields Without Graphs
arXiv:2509.24115v2 Announce Type: replace Abstract: Point defects play a central role in driving the properties of materials. First-principles methods are widely used to compute defect energetics and structures, including at scale for high-throughput defect databases. However, these methods are computationally expensive, making machine-learning force fields (MLFFs) an attractive alternative for accelerating structural relaxations. Most existing MLFFs are based on graph neural networks (GNNs), which can suffer from oversmoothing, oversquashing, and poor representation of long-range interactions. Both of these issues are especially of concern when modeling point defects. To address these challenges, we introduce the \textit{Accelerated Deep Atomic Potential Transformer} (ADAPT), an MLFF that replaces graph representations with a direct coordinates-in-space formulation and explicitly considers all pairwise atomic interactions. Atoms are treated as tokens, with a Transformer encoder modeling their interactions. Applied to a dataset of silicon point defects, ADAPT achieves a roughly 22% reduction in force and a roughly 40 percent reduction in energy prediction error relative to a state-of-the-art GNN-based model, while requiring only a fraction of the computational cost.
Abstract:Point defects play a central role in driving the properties of materials. First-principles methods are widely used to compute defect energetics and structures, including at scale for high-throughput defect databases. However, these methods are computationally expensive, making machine-learning force fields (MLFFs) an attractive alternative for accelerating structural relaxations. Most existing MLFFs are based on graph neural networks (GNNs), which can suffer from oversmoothing, oversquashing, and poor representation of long-range interactions. Both of these issues are especially of concern when modeling point defects. To address these challenges, we introduce the \textit{Accelerated Deep Atomic Potential Transformer} (ADAPT), an MLFF that replaces graph representations with a direct coordinates-in-space formulation and explicitly considers all pairwise atomic interactions. Atoms are treated as tokens, with a Transformer encoder modeling their interactions. Applied to a dataset of silicon point defects, ADAPT achieves a roughly 22% reduction in force and a roughly 40 percent reduction in energy prediction error relative to a state-of-the-art GNN-based model, while requiring only a fraction of the computational cost.
Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci); Optimization and Control (math.OC)
arXiv:2509.24115 [cs.LG]
(or arXiv:2509.24115v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2509.24115 Focus to learn more arXiv-issued DOI via DataCite
Phys. Rev. Research 8, 033266 (2026)
https://doi.org/10.1103/3xf3-ts5f Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 281 words
