Jul 2026· Proceedings of the Genetic and Evolutionary Computation Conference Companion· 0 citations· 25 references
TL;DR
A framework that combines large language models (LLMs) for problem understanding with a structured Biased Random-Key Genetic Algorithm (BRKGA) configurator for algorithm realization is presented, allowing users to describe optimization problems in natural language and receive executable GPU-accelerated GA implementations.
Abstract
Configuring genetic algorithms (GAs) for new optimization problems typically requires substantial expertise in representation design, decoder construction, and constraint handling. We present a framework that combines large language models (LLMs) for problem understanding with a structured Biased Random-Key Genetic Algorithm (BRKGA) configurator for algorithm realization, allowing users to describe optimization problems in natural language and receive executable GPU-accelerated GA implementations. We compare two workflows: LLM-Only, where the LLM generates a complete standalone GA from the problem description; and LLM+Config, where the LLM is restricted to problem parsing and decoder synthesis against a fixed brkga3 interface that provides the evolutionary architecture, Adaptive Large Neighborhood Search (ALNS) warm-start, GPU-parallel segmented sort, and bottleneck-focused local search. We evaluate on three problem families (the Traveling Salesperson Problem (TSP), the Traveling Salesperson Problem with Job-times (TSPJ), and the Vehicle Routing Problem with Resource Pick-up and Delivery (VRP-RPD)) on instances up to 999 nodes. LLM+Config outperforms or matches LLM-Only on every instance where both workflows produced a result, and additionally solves four TSP instances on which LLM-Only failed (Wilcoxon signed-rank with Holm correction across the three problem families, all adjusted p ≤ 0.012); on TSPJ it matches CPLEX optimal on 3 of 10 benchmarks and establishes first reported results for 6 large instances (n = 202–1,000); on VRP-RPD it wins all 13 instances against LLM-Only with 51–329 % advantage on large instances. GPU-parallel BRKGA evolution runs 25–35× faster per generation than a single-threaded CPU baseline. We characterise LLM stochasticity through two complementary studies: 15 sessions across 3 problems with one model (80% success rate), and 100 sessions across 10 LLMs from 3 providers on TSPJ/eil101 (Friedman and Wilcoxon-Holm tests on within-tier and within-provider blocks), finding that capability tier dominates over provider at the large tier and that small-tier non-Anthropic models contribute zero usable sessions.
LLM-driven evolutionary search can discover algorithm designs that achieve Pareto-efficient trade-offs difficult to reach through manual design, with SMAC hyperparameter optimization integrated into the evolutionary loop.
G. Laskaris, R. Brasher, Niki van Stein et al.· 0 citations
This work proposes a pipeline that introduces a novel regularization architecture balancing performance and complexity, and mitigate the side effects of automated tuning through two novel components: a symbolic pruning mutator and a complexity-aware mutation gate that explicitly filters out mutations leading to excessive code growth.
Guorui Quan, Mingfei Sun, Manuel López-Ibáñez et al.· Annual Conference on Genetic...· 0 citations
Evolutionary program search guided by Large Language Models (LLMs) has emerged as a powerful paradigm for automated scientific discovery. However, current approaches are fundamentally constrained by three bottlenecks: structurally blind parent selection, sparse whole-program evaluation rewards, and static mutation operators that fail to adapt during search. We present GAE (Graph-Augmented Evolution), a framework that resolves these limitations through a tightly coupled, three-pillar architecture. First, a relational graph neural network (GNN) parses programs into typed computation graphs, producing structure-aware embeddings. Second, an RL-optimized meta-controller leverages these embeddings to replace blind evolutionary sampling with a directed policy, dynamically selecting optimal parents and mutation directions based on reward history. Third, an online GRPO fine-tuning loop continuously updates the LLM mutation operator at test-time using group-normalized evaluation rewards, directly aligning the model's generation distribution with high-fitness structural edits. We evaluate GAE on a challenging scientific discovery task: symbolic regression for complex nonlinear oscillator systems. By transforming stochastic search into a directed, self-improving trajectory, GAE efficiently discovers closed-form physical equations, consistently matching or outperforming static LLM-driven baselines and achieving state-of-the-art out-of-distribution performance.
This work proposes a general LES framework that incorporates feature-assisted niche construction within abstract search spaces, enabling the seamless integration of niche-based search strategies from evolutionary computation and introduces PartEvo (Partition to Evolve), an LES method that combines niche collaborative search and advanced prompting strategies to improve algorithm discovery efficiency.
Qinglong Hu, Qingfu Zhang· Neural Information Processin...· 10 citations· ⚡3
Combinatorial problems appear in numerous industrial applications. A common approach is to formulate these problems as declarative constraint models that can subsequently be compiled to and solved by a range of back-end solvers. Recent work shows that Large Language Models (LLMs) can produce correct models from natural language, but even a correct model can be expensive to solve because performance remains sensitive to modelling choices. In this work, we investigate whether LLMs can automate performance-oriented model reformulation. Inspired by Automatic Heuristic Design (AHD), we use an evolutionary framework in which an LLM proposes candidate reformulations that are verified and benchmarked against the user-defined baseline model. We compare AHD-adapted search strategies that control which prior attempts, instructions, and measured feedback enter each prompt. Existing retention strategies prioritize recency or performance, but do not explicitly diversify the context. To cover this gap, we introduce Profile-Diverse Retention (PDR), which applies Maximal Marginal Relevance (MMR) to instance-level runtime vectors to retain behaviourally diverse attempts. We systematically evaluate the strategies on eight CSPLib problems using validation-based final model selection. The results show that: (i) iterative reformulation can produce substantial held-out speedups; (ii) strategies that keep the retained context diverse outperform those that retain only recent or the fastest attempts; and (iii) validation-based selection improves the held-out speedup of every strategy.
Kostis Michailidis, Dimos Tsouros, Nguyen Dang et al.· 0 citations
The Traveling Salesman Problem is a canonical setting for studying how a population-based method should allocate a fixed search budget between exploration and progressively stronger local intensification. We propose ADPSO-ERLS, a discrete swarm algorithm that treats this allocation as an explicit, tunable design variable. It couples memory-guided swap mutation, heterogeneous initialization, selective candidate-restricted 2-opt during evolution, and an incumbent-only final refinement combining candidate-restricted and optional full 2-opt with double-bridge perturbations. The method is PSO-inspired, using personal and global memories yet dispensing with velocity, inertia, and acceleration coefficients. All six algorithms are implemented in Rust, run on identical hardware, and stopped at a strict, recorded limit of 100,000 candidate-solution assessments, so that programming language, hardware, and evaluation budget are held common across methods; wall-clock time is reported separately because equal assessment counts need not correspond to equal arithmetic work. Over 50 runs on five symmetric TSPLIB instances under the integer \texttt{EUC\_2D} convention, ADPSO-ERLS attains the lowest best and mean cost on every instance, with best-tour Gap of $1.93$--$4.17\%$ and relative error of $3.04$--$5.50\%$. It ranks first under the Friedman test, and all twenty-five multiplicity-controlled Wilcoxon comparisons favor it with large, near-complete distributional separation. A paired ablation with common seeds links initialization, in-run local search, and final refinement to quality gains, while candidate restriction chiefly cuts runtime, by up to a factor of roughly $38$. Further experiments up to $16{,}862$ cities keep best-tour Gaps below $6.7\%$, solving the largest case in under eleven minutes.
A. Soria-Lorente, Jean-Marie Vilaire, Junior Michel et al.· 0 citations