{"ID":2866247,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.00027","arxiv_id":"2510.00027","title":"Learning Inter-Atomic Potentials without Explicit Equivariance","abstract":"Accurate and scalable machine-learned inter-atomic potentials (MLIPs) are essential for molecular simulations ranging from drug discovery to new material design. Current state-of-the-art models enforce roto-translational symmetries through equivariant neural network architectures, a hard-wired inductive bias that can often lead to reduced flexibility, computational efficiency, and scalability. In this work, we introduce TransIP: Transformer-based Inter-Atomic Potentials, a novel training paradigm for interatomic potentials achieving symmetry compliance without explicit architectural constraints. Our approach guides a generic non-equivariant Transformer-based model to learn SO(3)-equivariance by optimizing its representations in the embedding space. Trained on the recent Open Molecules (OMol25) collection, a large and diverse molecular dataset built specifically for MLIPs and covering different types of molecules (including small organics, biomolecular fragments, and electrolyte-like species), TransIP attains comparable performance in machine-learning force fields versus state-of-the-art equivariant baselines. Further, compared to a data augmentation baseline, TransIP achieves 40% to 60% improvement in performance across varying OMol25 dataset sizes. More broadly, our work shows that learned equivariance can be a powerful and efficient alternative to equivariant or augmentation-based MLIP models. Our code is available at: https://github.com/Ahmed-A-A-Elhag/TransIP.","short_abstract":"Accurate and scalable machine-learned inter-atomic potentials (MLIPs) are essential for molecular simulations ranging from drug discovery to new material design. Current state-of-the-art models enforce roto-translational symmetries through equivariant neural network architectures, a hard-wired inductive bias that can o...","url_abs":"https://arxiv.org/abs/2510.00027","url_pdf":"https://arxiv.org/pdf/2510.00027v3","authors":"[\"Ahmed A. Elhag\",\"Arun Raja\",\"Alex Morehead\",\"Samuel M. Blau\",\"Hongtao Zhao\",\"Christian Tyrchan\",\"Eva Nittinger\",\"Garrett M. Morris\",\"Michael M. Bronstein\"]","published":"2025-09-25T22:15:10Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.AI\",\"q-bio.BM\",\"q-bio.QM\"]","methods":"[\"Transformer\",\"Generative Adversarial Network\"]","has_code":false,"code_links":[{"ID":609354,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2866247,"paper_url":"https://arxiv.org/abs/2510.00027","paper_title":"Learning Inter-Atomic Potentials without Explicit Equivariance","repo_url":"https://github.com/Ahmed-A-A-Elhag/TransIP","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}