Algorithmic algorithm development with LLMs: A Case Study on LLM-Usage for Contraction Order Optimization in Tensor Networks
Fabian Hoppe, Melven R\"ohrig-Z\"ollner, Philipp Knechtges
Why It Matters
What makes this one worth your time
Understanding the role of LLMs in algorithm development can enhance efficiency in computational tasks, particularly in complex areas like tensor networks.
This study explores LLMs in algorithm development for tensor networks, emphasizing human oversight.
Summary
The paper presents a case study on using LLMs for developing algorithms, specifically focusing on contraction order optimization in tensor networks, while discussing the importance of evaluation and validation by human scientists.
Key contributions
- Case study on LLM usage for contraction order optimization in tensor networks.
- Analysis of design choices in LLM-based algorithm development.
- Discussion on the challenges of human evaluation and interpretation in the process.
Notable insights
- The study highlights the significance of choosing the right LLM and evaluation metrics in algorithm development.
- It suggests that verifier-guided evolutionary coding agents can aid in improving algorithm performance.
Possible limitations
- Not stated in the abstract.
Abstract
arXiv:2606.01975v1 Announce Type: new Abstract: We consider LLM-based algorithm development through a case study on contractionorder optimisation for tensor networks with OpenEvolve. We pay particular attention to the choice of the LLM as well as design choices such as evaluation metric and test instances. Our results highlight both the promise of verifier-guided evolutionary coding agents for algorithm development/improvement and the continuing importance of evaluation, validation, and interpretation -- and corresponding challenges -- by the human scientist.