Automated Large-scale CVRP Solver Design via LLM-assisted Flexible MCTS

Tong Guo, Caishun Chen, Yew Soon Ong · 2026

Read Automated Large-scale CVRP Solver Design via LLM-assisted Flexible MCTS by Tong Guo, Caishun Chen, Yew Soon Ong (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-2605-03339.txt.

Description

Tong Guo, Caishun Chen, and Yew Soon Ong automate the design of solvers for very large vehicle routing problems, where hundreds to thousands of stops defeat even strong solvers. Splitting the problem helps, but crafting the split logic and tuning sub-solvers takes deep expert labor.

Their LaF-MCTS framework uses language models inside a flexible tree search, organized in three decision tiers for split policies and sub-solvers. Semantic pruning drops redundant code, while branch regrowth restores diversity. On CVRPLib tests, the system built split-enhanced solvers that beat several state-of-the-art routing solvers without hand design.

Catalog record

CREATORTong Guo, Caishun Chen, Yew Soon Ong
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.03339

Browse the full library · About the Tower of Babel