arXiv：cs.LG· Michelangelo Domina, Michele Ceriotti·· 5 小时前AI 评分37
利用大语言模型探索原子中心结构描述符的极限
Using large language models to probe the limits of atom-centered structural descriptors
AI 导读
研究者借助大语言模型发现了即使考虑多达七个邻居的团簇、乃至在实用离散化水平下任意阶数仍无法区分的三维结构，突破了此前所有"简并"都能被更大邻居团簇化解的认知。这些结构的关键构造要素可追溯至不同领域数十年前已知的结果，模型成功找到了相关文献并识别出其对当前问题的意义。该实验揭示了一种AI用于科学的高效模式：在不同领域间迁移成果，加速一个领域的偶然发现成为另一领域突破的进程。
正文
Abstract:Mapping an atomic structure to a compact set of geometric descriptors is an essential step in any machine-learning application to atomic-scale modeling. A powerful and widely-used approach can be understood as a discretization of the histogram of pair distances, triangles, etc., that results in a hierarchy of symmetry-invariant atom-centered descriptors. Unfortunately, the lower rungs on this hierarchy (two, three, four-neighbor clusters) were found to be incomplete, with symmetry-unrelated pairs of structures having exactly the same descriptors. However, all the ``degeneracies'' reported so far are resolved by considering larger clusters of neighbors to build the descriptors. We report examples of 3D structures that are indistinguishable even if one considers clusters of up to seven neighbors, and to arbitrary order when considering a practical level of discretization of the descriptors, discovered with the assistance of large language models. The key ingredients in their construction can be traced to results that have been known for decades in different communities: the model was able to find the references and recognize their significance for the problem at hand. We believe this experiment exposes an extremely fruitful usage pattern for AI in science: translating results between different communities and application domains, accelerating the process by which serendipitous discoveries in a field become breakthroughs in another.
| Subjects: | Chemical Physics (physics.chem-ph); Machine Learning (cs.LG) |
| Cite as: | arXiv:2607.26984 [physics.chem-ph] |
| (or arXiv:2607.26984v2 [physics.chem-ph] for this version) | |
| https://doi.org/10.48550/arXiv.2607.26984 arXiv-issued DOI via DataCite |
Submission history
From: Michelangelo Domina [view email]
[v1]
Wed, 29 Jul 2026 14:40:59 UTC (2,228 KB)
[v2]
Fri, 2 Oct 2026 16:28:51 UTC (2,223 KB)
来源：arXiv：cs.LG · arxiv.org