# Early Discoveries of Algorithmist I: Promise of Provable Algorithm Synthesis at Scale

arXiv: https://arxiv.org/abs/2603.22363 · HTML: https://mlsanity.com/paper/2603.22363/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Mathematical reasoning, one of 17 directions in LLM reasoning.
- Builds on 22 of its 119 references that are in the map.
- First posted on arXiv 23 Mar 2026.

## Placed in

- LLM reasoning / [Mathematical reasoning](https://mlsanity.com/lr/mathematical-reasoning/)

## Builds on (most cited first, 22 of 22)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [AlphaEvolve: A coding agent for scientific and algorithmic discovery](https://mlsanity.com/paper/2506.13131/) (2025)
- [Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents](https://mlsanity.com/paper/2505.22954/) (2025)
- [DeepSeek-Prover-V2: Advancing Formal Mathematical Reasoning via Reinforcement Learning for Subgoal Decomposition](https://mlsanity.com/paper/2504.21801/) (2025)
- [Goedel-Prover-V2: Scaling Formal Theorem Proving with Scaffolded Data Synthesis and Self-Correction](https://mlsanity.com/paper/2508.03613/) (2025)
- [DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning](https://mlsanity.com/paper/2511.22570/) (2025)
- [A Self-Improving Coding Agent](https://mlsanity.com/paper/2504.15228/) (2025)
- [Lean Copilot: Large Language Models as Copilots for Theorem Proving in Lean](https://mlsanity.com/paper/2404.12534/) (2024)
- [Aristotle: IMO-level Automated Theorem Proving](https://mlsanity.com/paper/2510.01346/) (2025)
- [You Name It, I Run It: An LLM Agent to Execute Tests of Arbitrary Projects](https://mlsanity.com/paper/2412.10133/) (2024)
- [Early science acceleration experiments with GPT-5](https://mlsanity.com/paper/2511.16072/) (2025)
- [ThunderKittens: Simple, Fast, and Adorable AI Kernels](https://mlsanity.com/paper/2410.20399/) (2024)
- [AgentCoder: Multi-Agent-based Code Generation with Iterative Testing and Optimisation](https://mlsanity.com/paper/2312.13010/) (2023)
- [Accelerating Scientific Research with Gemini: Case Studies and Common Techniques](https://mlsanity.com/paper/2602.03837/) (2026)
- [Differentially private k-means clustering via exponential mechanism and max cover](https://mlsanity.com/paper/2009.01220/) (2020)
- [Differentially private partition selection](https://mlsanity.com/paper/2006.03684/) (2020)
- [The Price of Explainability for Clustering](https://mlsanity.com/paper/2304.09743/) (2023)
- [KernelEvolve: Scaling Agentic Kernel Coding for Heterogeneous AI Accelerators at Meta](https://mlsanity.com/paper/2512.23236/) (2025)
- [Helix: Evolutionary Reinforcement Learning for Open-Ended Scientific Problem Solving](https://mlsanity.com/paper/2603.07642/) (2026)
- [Dynamic Algorithm for Explainable k-medians Clustering under lp Norm](https://mlsanity.com/paper/2512.01150/) (2025)

## Built on by (most cited first, 1 of 1)

- [CliffSearch: Structured Agentic Co-Evolution over Theory and Code for Scientific Algorithm Discovery](https://mlsanity.com/paper/2604.01210/) (2026)

## Abstract (arXiv)

Designing algorithms with provable guarantees that also work well in practice remains difficult, requiring both mathematical reasoning and careful implementation. Existing approaches that bridge worst-case theory and empirical performance, such as beyond-worst-case analysis and data-driven algorithm selection, typically assume prior distributional knowledge or restrict attention to a fixed pool of algorithms. Recent progress in LLMs suggests a new possibility: provable algorithm synthesis on the fly. To study this, we built Algorithmist, an autonomous researcher agent on top of GitHub Copilot that runs a multi-agent research-and-review loop, with separate stages for idea generation, algorithm and proof development, proof-guided implementation, and review of proofs, code, and their alignment. We evaluate Algorithmist on research-level tasks in private data analysis and clustering. When asked to design practical methods that jointly satisfy privacy, approximation, and interpretability requirements, it produced provably sound and empirically effective algorithms, together with research-style writeups and audited implementations. It also found improved algorithms in some settings, explained principled barriers in others, and uncovered a subtle proof bug in prior published work. More broadly, our results suggest a new paradigm in which LLM systems generate research-paper-quality algorithmic artifacts tailored to each dataset and deployment setting. They also point to a proof-first code-synthesis paradigm, in which code is developed alongside a structured natural-language proof intermediate representation and kept aligned with it throughout synthesis.