RamanGPT: Bidirectional Mapping Between Crystal Structures and Raman Spectra with Graph Neural Networks and Generative Transformers
Frank M. Abel, Jaehyung Lee, Charles R. Campbell, Kamal Choudhary · 2026
Read RamanGPT: Bidirectional Mapping Between Crystal Structures and Raman Spectra with Graph Neural Networks and Generative Transformers by Frank M. Abel, Jaehyung Lee, Charles R. Campbell, Kamal Choudhary (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-03764.txt.
Description
Abel, Lee, Campbell, and Choudhary map crystal structures to Raman spectra and back with RamanGPT. Raman spectroscopy is a handy vibrational probe in materials labs. Predicting a spectrum from a structure usually needs costly perturbation theory. Reading a structure from a spectrum usually needs curated reference matching. Their framework tackles both directions for inorganic crystals.
A line graph neural network predicts spectra from structures. It trains on a database of about 5,100 materials. Output covers 200 bins across the standard wavenumber range. Many predictions capture the target spectrum's main features. A tuned language model runs the inverse direction from spectrum plus formula prompts. It recovers lattice values with small errors on held-out materials. Formula consistency reaches about 87 percent. A matcher and a consistency loop are live on a public site. Materials teams get fast two-way Raman tools.
Catalog record
| CREATOR | Frank M. Abel, Jaehyung Lee, Charles R. Campbell, Kamal Choudhary |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.03764