Solver-Independent Automated Problem Formulation via LLMs for High-Cost Simulation-Driven Design
Yuchen Li, Handing Wang, Bing Xue, Mengjie Zhang, Yaochu Jin · 2025
Read Solver-Independent Automated Problem Formulation via LLMs for High-Cost Simulation-Driven Design by Yuchen Li, Handing Wang, Bing Xue, Mengjie Zhang, Yaochu Jin (2025) 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-2512-18682.txt.
Description
Yuchen Li and four coauthors tackle a slow step in engineering design in this 2025 paper. When simulations are costly, turning loose design wishes into a precise optimization problem is done by hand and eats expert time. Language models could automate the step, but earlier attempts either drifted from the design intent or needed solver feedback to filter training data, feedback that costly simulations cannot supply. Their framework, APF, works without any solver in the loop.
APF builds its own fine-tuning data through a generation and annotation pipeline, then trains models to output accurate, runnable optimization models. On antenna design tasks it beats existing methods both in how faithfully requirements are formalized and in the quality of the radiation efficiency curves produced.
Catalog record
| CREATOR | Yuchen Li, Handing Wang, Bing Xue, Mengjie Zhang, Yaochu Jin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 48 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-18682