A knowledge-first approach to LLM-driven automatic heuristic design in combinatorial optimization yields better discovery efficiency, transfer, and generalization than code-centric baselines by formalizing a distortion-compression trade-off.
Algorithm evolution using large language model
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
verdicts
UNVERDICTED 5representative citing papers
LLMs optimize genetic algorithm operations and create a new heuristic AutoPort for port selection and beamforming in fluid antenna systems to maximize min SINR, achieving near-optimal performance in simulations.
MEP uses LLMs in a structured reasoning cycle to evolve improved heuristics for HGS on VRPs, achieving up to 2.7% better solution quality and over 45% reduced runtime.
Fine-tuned LLMs with DAR sampling and DPO outperform off-the-shelf versions on algorithm design tasks and generalize to related settings.
Under fixed token budget on Circle Packing, deeper per-candidate reasoning beats generating more shallow candidates, and capable models produce evaluation hacks at higher rates.
citing papers explorer
-
Back to the Beginning of Heuristic Design: Bridging Code and Knowledge with LLMs
A knowledge-first approach to LLM-driven automatic heuristic design in combinatorial optimization yields better discovery efficiency, transfer, and generalization than code-centric baselines by formalizing a distortion-compression trade-off.
-
LLM-Enabled Automated Algorithm Design for Multiuser Fluid Antenna Communications
LLMs optimize genetic algorithm operations and create a new heuristic AutoPort for port selection and beamforming in fluid antenna systems to maximize min SINR, achieving near-optimal performance in simulations.
-
PyVRP$^+$: LLM-Driven Metacognitive Heuristic Evolution for Hybrid Genetic Search in Vehicle Routing Problems
MEP uses LLMs in a structured reasoning cycle to evolve improved heuristics for HGS on VRPs, achieving up to 2.7% better solution quality and over 45% reduced runtime.
-
Effective Harness Engineering for Algorithm Discovery with Coding Agents
Under fixed token budget on Circle Packing, deeper per-candidate reasoning beats generating more shallow candidates, and capable models produce evaluation hacks at higher rates.