arXiv cs.AI· Yingheng Wang, Tao Yu, Shufeng Kong, Francesco Ricci, John M. Gregoire, Carla P. Gomes·· 3 小时前AI 评分38
DeepSciReasoner：物理约束下的深度科学推理与结构感知光谱预测
Deep Scientific Reasoning under Physical Constraints: Structure-Aware Spectrum Prediction
AI 导读
研究团队提出 DeepSciReasoner，一个在物理约束下预测科学光谱的通用框架，结合结构感知光谱解码与保约束物理推理，以电子态密度（eDOS）为典型示例。该框架在 eDOS、声子态密度（phDOS）、XANES 和 Raman 光谱上显著提升预测精度并保持物理一致性，代码与数据已在项目页面开源，成果发表于 ICML 2026。
正文 · 原文
Abstract:Structured scientific spectra encode rich physical information while obeying hard constraints, such as conservation laws and characteristic spectral geometry. Accurate prediction of these spectra is central to materials discovery, yet existing methods often treat them as unconstrained sequences and therefore fail to enforce the underlying physical structure. Taking electronic density of states (eDOS) as the prototypical example, we introduce DeepSciReasoner, a general paradigm for predicting scientific spectra under physical constraints. The framework combines structure-aware spectrum decoding with constraint-preserving physical reasoning, allowing predictions to capture rich spectral structure while respecting the underlying physics. We evaluate DeepSciReasoner on eDOS, phonon density of states (phDOS), X-ray absorption near-edge structure (XANES), and Raman spectra, where it substantially improves prediction accuracy while maintaining physical consistency. These results establish DeepSciReasoner as a reusable blueprint for structured scientific spectrum prediction under hard physical constraints.
| Comments: | Published in ICML 2026. Code and data are available on the project page: this https URL |
| Subjects: | Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.34021 [cond-mat.mtrl-sci] |
| (or arXiv:2610.34021v1 [cond-mat.mtrl-sci] for this version) | |
| https://doi.org/10.48550/arXiv.2610.34021 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yingheng Wang [view email]
[v1]
Sun, 25 Oct 2026 03:22:14 UTC (547 KB)
来源：arXiv cs.AI · arxiv.org