The paper, titled “Prompt-Augmentation for Evolving Heuristics for a Niche Optimization Problem,” explores the use of Large Language Models (LLMs) for the automated development of heuristics for specialized optimization problems. It introduces two newly developed approaches: Contextual Evolution of Heuristics (CEoH), which integrates problem-specific information into prompt design, and Literature-Based CEoH (LitCEoH), which additionally incorporates knowledge from scientific literature. The goal is to enable LLMs to design effective and generalizable heuristics.
The results show that these strategies allow smaller, energy-efficient language models to generate high-quality heuristics—sometimes outperforming much larger models. This work makes an important contribution to advancing automated, AI-driven heuristic design in optimization research.