从手工设计到基于大语言模型的元启发式变异算子：教程

中文摘要 AI 辅助

大语言模型（LLMs）正越来越多地被用作元启发式算法中的变异算子，在迭代搜索循环内生成或修改候选解、启发式规则或程序。这一转变将变异重新定义为一种模型调用，其条件依赖于不同类型的信息。我们引入了一个算子级框架，包含两个描述符：（1）变异时提示条件信息的类型（\texttt{Numeric}数值型、\texttt{Symbolic}符号型、\texttt{Linguistic}语言型），以及（2）工件持久性，即识别在模型调用后仍保留的内容（\texttt{Transient}瞬时型、\texttt{Amortized}摊销型、\texttt{Transfer}迁移型）。本教程通过一个可操作的构建模板、方法综述、证据表格和成本感知决策指南，展示了如何分类、构建和选择这些算子。

英文摘要

Large language models (LLMs) are increasingly being employed as variation operators in metaheuristics, generating or modifying candidate solutions, heuristics, or programs inside iterative search loops. This shift reframes variation as a model call conditioned on different types of information. We introduce an operator-level framework with two descriptors: (1) the type of prompt-conditioning information at variation time (\texttt{Numeric}, \texttt{Symbolic}, \texttt{Linguistic}), and (2) artifact persistence, identifying what survives the model call (\texttt{Transient}, \texttt{Amortized}, \texttt{Transfer}). The tutorial shows how to classify, build, and select these operators through a worked build template, a method survey, an evidence table, and a cost-aware decision guide.

发表机构

• Apeiron Intelligence（阿派朗智能）
• Artificial Intelligence Research Institute (IIIA-CSIC)（人工智能研究所（IIIA-CSIC））
• Universitat Politècnica de Catalunya (UPC)（加泰罗尼亚理工大学（UPC））

机构由 AI 辅助整理，请以论文原文为准。