LLM-Driven Co-Evolutionary Automated Heuristic Design for Bi-Component Coupled Combinatorial Optimization
Mingen Kuang, Xudong Deng, Xi Lin, Ye Fan, Jianyong Sun, Jialong Shi · 2026
Read LLM-Driven Co-Evolutionary Automated Heuristic Design for Bi-Component Coupled Combinatorial Optimization by Mingen Kuang, Xudong Deng, Xi Lin, Ye Fan, Jianyong Sun, Jialong Shi (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-2606-00718.txt.
Description
Kuang and co-authors use language models to design search rules in this 2026 paper. Their field is automated heuristic design, where software writes the rules that solve hard routing and choice tasks. Past tools grow one rule at a time. That misses tasks where two choices lock tight, like the Traveling Thief Problem.
Their system, CoEvo-AHD, grows two groups of rules side by side. One group handles routes, and one handles item choice. A joint scoring step rewards pairs that work well as a team. A shared tool library gives the rules ready-made parts, like fast local-search math, through clean call points. On the Traveling Thief and Traveling Purchaser tasks, the system finds rule pairs that team up well. The authors report results that rival classic hand-built heuristics.
Catalog record
| CREATOR | Mingen Kuang, Xudong Deng, Xi Lin, Ye Fan, Jianyong Sun, Jialong Shi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 84 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.00718