Abstract
Large language models are increasingly participating in complex real-world tasks in the form of algorithm-design agents, designing and refining algorithms. Many successful algorithm-design agents adopt pure in-context evolutionary frameworks, but they may quickly plateau in domains that require specialized knowledge. Parametric adaptation offers a way to internalize specialized knowledge, but conventional training requires abundant domain-specific corpora while high-quality algorithms are scarce in complex algorithm-design scenarios. In this paper, we propose sample-efficient parametric self-evolution where agents can explore and learn from self-generated algorithms. First, we characterize in-context evolutionary stagnation and analytically propose the Improvement Chain proposition, showing how learning successive self-generated algorithms can locally increase the likelihood of neighboring algorithms. Motivated by this local-transfer perspective, we further propose Population-Curated Policy Optimization (PCPO) to utilize a global population and a hybrid policy update scheme for retaining and reusing high-quality, diverse self-generated algorithms, shifting the policy towards stronger algorithms. In the task of learning rate schedule design for global placement in electronic design automation, trained only on 4 chip cases, PCPO outperforms the state-of-the-art in-context evolutionary methods (e.g., OpenEvolve and ShinkaEvolve) on average across 16 chip cases. With an 8B-size base model, PCPO achieves competitive performance compared to frontier closed-source models such as GPT-5.5. PCPO also reduces inference-time token cost by internalizing grounded domain knowledge and prompt distillation. Moreover, PCPO achieves significant speedups on four GPU kernel designs, with an average of 8.27$\times$ speedup against the PyTorch Eager baseline.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Lu, C., Xue, K., Xu, S., Yuan, M., & Qian, C. (2026). Self-Evolving Algorithm-Design Agents: Escaping In-Context Evolutionary Stagnation via Population-Curated Policy Optimization. https://omanscience.com/en/articles/self-evolving-algorithm-design-agents-escaping-in-context-evolutionary-stagnation-via-population-curated-policy-optimization
MLA 9
Lu, Chen, et al. "Self-Evolving Algorithm-Design Agents: Escaping In-Context Evolutionary Stagnation via Population-Curated Policy Optimization." https://omanscience.com/en/articles/self-evolving-algorithm-design-agents-escaping-in-context-evolutionary-stagnation-via-population-curated-policy-optimization.
Chicago (author–date)
Lu, Chen, Ke Xue, Siyuan Xu, Mingxuan Yuan, and Chao Qian. 2026. "Self-Evolving Algorithm-Design Agents: Escaping In-Context Evolutionary Stagnation via Population-Curated Policy Optimization." https://omanscience.com/en/articles/self-evolving-algorithm-design-agents-escaping-in-context-evolutionary-stagnation-via-population-curated-policy-optimization.
Harvard
Lu, C., Xue, K., Xu, S., Yuan, M. and Qian, C. (2026) 'Self-Evolving Algorithm-Design Agents: Escaping In-Context Evolutionary Stagnation via Population-Curated Policy Optimization', Available at: https://omanscience.com/en/articles/self-evolving-algorithm-design-agents-escaping-in-context-evolutionary-stagnation-via-population-curated-policy-optimization.
Vancouver
Lu C, Xue K, Xu S, Yuan M, Qian C. Self-Evolving Algorithm-Design Agents: Escaping In-Context Evolutionary Stagnation via Population-Curated Policy Optimization. https://omanscience.com/en/articles/self-evolving-algorithm-design-agents-escaping-in-context-evolutionary-stagnation-via-population-curated-policy-optimization
IEEE
C. Lu, K. Xue, S. Xu, M. Yuan, and C. Qian, "Self-Evolving Algorithm-Design Agents: Escaping In-Context Evolutionary Stagnation via Population-Curated Policy Optimization," https://omanscience.com/en/articles/self-evolving-algorithm-design-agents-escaping-in-context-evolutionary-stagnation-via-population-curated-policy-optimization.