DeltaEvolve: Accelerating Scientific Discovery through Momentum-Driven Evolution

2026-02-02

AI Summary

This work proposes a momentum-driven expectation-maximization evolutionary framework that addresses the limitations of existing large language model–based evolutionary systems, which rely on full code histories and consequently suffer from contextual redundancy and weak evolutionary signals. Instead of retaining complete snapshots, the proposed method leverages structured semantic increments to represent program evolution. By organizing these increments in a multi-level database and employing a progressive context disclosure mechanism, the framework efficiently captures critical improvements across evolutionary trajectories. This approach significantly enhances both the efficiency and guidance of the evolutionary process, consistently yielding superior solutions with fewer token expenditures across multiple scientific discovery tasks compared to conventional history-based evolutionary methods.

Technology Category

• Machine Learning: Evolutionary Learning
• Search and Optimization: Evolutionary Computation
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Models for Web evolution
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

LLM-driven evolutionary systems have shown promise for automated science discovery, yet existing approaches such as AlphaEvolve rely on full-code histories that are context-inefficient and potentially provide weak evolutionary guidance. In this work, we first formalize the evolutionary agents as a general Expectation-Maximization framework, where the language model samples candidate programs (E-step) and the system updates the control context based on evaluation feedback (M-step). Under this view, constructing context via full-code snapshots constitutes a suboptimal M-step, as redundant implement details dilutes core algorithmic ideas, making it difficult to provide clear inspirations for evolution. To address this, we propose DeltaEvolve, a momentum-driven evolutionary framework that replaces full-code history with structured semantic delta capturing how and why modifications between successive nodes affect performance. As programs are often decomposable, semantic delta usually contains many effective components which are transferable and more informative to drive improvement. By organizing semantic delta through multi-level database and progressive disclosure mechanism, input tokens are further reduced. Empirical evaluations on tasks across diverse scientific domains show that our framework can discover better solution with less token consumption over full-code-based evolutionary agents.

Problem

• evolutionary agents
• code history
• context efficiency
• scientific discovery
• semantic delta

Innovation

• semantic delta
• momentum-driven evolution
• Expectation-Maximization framework
• LLM-driven scientific discovery
• context efficiency