Deep Scientific Reasoning under Physical Constraints: Structure-Aware Spectrum Prediction
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.
Related
- Prospector: learning a compact graph traversal policy for scientific reasoning
- On physical implementability of reinforcement-learning-driven reasoning
- Physical Law Ecology: mapping multi-mechanism ecologies as the zeroth step of data-driven scientific discovery
- Constraints for Physical Gauge Groups Coming from the Causal Action Principle
- Deformation Quantization of Schutz Quantum Cosmology: Relational Time, Constraints, and Physical States
BibTeX
@misc{wang2026deep,
title = {{Deep Scientific Reasoning under Physical Constraints: Structure-Aware Spectrum Prediction}},
author = {Yingheng Wang and Tao Yu and Shufeng Kong and Francesco Ricci and John M. Gregoire and Carla P. Gomes},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.34021},
eprint = {2610.34021},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.34021},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Deep Scientific Reasoning under Physical Constraints: Structure-Aware Spectrum Prediction
AU - Yingheng Wang
AU - Tao Yu
AU - Shufeng Kong
AU - Francesco Ricci
AU - John M. Gregoire
AU - Carla P. Gomes
PY - 2026
DA - 2026/10/25
PB - arXiv
DO - 10.48550/arxiv.2610.34021
UR - https://arxiv.org/abs/2610.34021
AB - 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.
ER -