EvoMem: Memory-Augmented Evolution for Code Optimization

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Viktor Volkov, Valentin Khrulkov, Andrey V. Galichin, Danil Sivtsov, Nikita Glazkov, Olga Volkova, Konstantin Pchelin, Iaroslav Bespalov, Dmitry V. Dylov, Petr Anokhin, Ivan Oseledets
• Venue: arXiv: cs.AI
• Published: 2026-08-11
• Collected: 2026-08-12
• URL: http://arxiv.org/abs/2608.10795v1
• PDF: https://arxiv.org/pdf/2608.10795v1
• DOI: UNKNOWN
• Registry group: academic_apis_and_preprints
• Evidence role: canonical_academic_index
• Language: en
• Country / region: UNKNOWN

Plain-Language Brief

The paper introduces EvoMem, a memory architecture for Long Learning Machine (LLM) based evolutionary program search. EvoMem captures and reuses knowledge from successful mutation events to improve the efficiency of future runs across various tasks.

What the Document Says

• EvoMem is a persistent memory architecture for LLM-based evolutionary program search.
• It stores promising ideas from successful mutation events with their provenance.
• During subsequent evolution, it retrieves relevant instructions based on the current task and program context to guide mutation.
• The paper reports positive improvements in target metrics or search speed for most evaluated settings across geometric optimization, multi-hop question answering, GPU kernel optimization, and related benchmarks.

Key Concepts Mentioned

• EvoMem
• Long Learning Machine (LLM)
• Evolutionary program search
• Mutation events
• Geometric optimization
• Multi-hop question answering
• GPU kernel optimization

Claims Made by the Document

• Existing LLM-driven evolutionary frameworks largely discard reusable knowledge, limiting opportunities for cross-run and cross-task learning.
• EvoMem reduces some redundant exploration and improves the reuse and adaptation of successful strategies in LLM-driven evolutionary search.

Evidence Used

• The text mentions that EvoMem converts successful mutation events into structured advice for future runs.
• The text reports positive improvements across various tasks.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Successful mutation strategies in evolutionary code search may contain reusable knowledge that is useful beyond a single run, and in some cases may transfer across related tasks and domains. However, existing LLM-driven evolutionary frameworks largely discard such knowledge, repeatedly rediscovering similar ideas and limiting opportunities for cross-run and cross-task learning. We introduce EvoMem, a persistent memory architecture for LLM-based evolutionary program search that captures and reuses candidate mutation knowledge. EvoMem converts successful mutation events into structured, task-aware advice for future runs. It operates in two phases: after each run, it extracts and stores promising ideas with provenance, and during subsequent evolution, it retrieves a small set of relevant instructions based on the current task and program context to guide mutation. Across geometric optimization, multi-hop question answering, GPU kernel optimization, and related benchmarks, our experiments show positive average improvements in target metrics or search speed for most evaluated settings, while also revealing variability across tasks. Overall, EvoMem provides evidence that persistent memory can reduce some redundant exploration and improve the reuse and adaptation of successful strategies in LLM-driven evolutionary search.