# From LLM-Generated Conjectures to Lean Formalizations: Automated Polynomial Inequality Proving via Sum-of-Squares Certificates

Published: 2026-05-19 · Updated: 2026-05-19 · Category: Other

Source page: https://www.sipply.dev/story/from-llm-generated-conjectures-to-lean-formalizations-701f3c64 · 中文: https://www.sipply.dev/zh/story/from-llm-generated-conjectures-to-lean-formalizations-701f3c64

## Summary

Recent advancements in the intersection of artificial intelligence and formal mathematics have been highlighted by the introduction of NSPI, a neuro-symbolic framework that combines large language models (LLMs) with symbolic computation. This innovative approach aims to automate the proving of polynomial inequalities, a task that has traditionally required significant human expertise. By leveraging machine-checked proofs, NSPI not only enhances the scalability of these proofs but also bridges the gap between conjectures generated by LLMs and formal mathematical verification using the Lean proof assistant.

## Key takeaways

- NSPI integrates LLMs with symbolic computation for polynomial inequality proving.
- The framework enhances scalability through machine-checked proofs.
- It represents a significant step towards automating formal mathematical verification.

## Why it matters

This development signifies a major leap in automating complex mathematical proofs, which can lead to more efficient verification processes in various scientific and engineering fields.

- **Product managers:** For product managers, understanding NSPI's capabilities can inform the development of tools that leverage AI for formal verification, potentially reducing time and resources spent on complex mathematical tasks.
- **Developers:** Developers and engineers can explore NSPI as a new tool for integrating AI into their workflows, particularly in areas requiring rigorous mathematical proofs and formal verification.
- **Students and job seekers:** Students and job-seekers in AI and mathematics should be aware of NSPI as it represents a growing field where expertise in both AI and formal methods can lead to innovative career opportunities.

## Sources

- [From LLM-Generated Conjectures to Lean Formalizations: Automated Polynomial Inequality Proving via Sum-of-Squares Certificates](https://arxiv.org/abs/2605.15445) (arXiv cs.AI, 2026-05-19)

## 中文

### 从LLM生成的猜想到Lean形式化：通过平方和证书自动化多项式不等式证明

最近在人工智能与形式数学交叉领域的进展通过NSPI得到了体现，NSPI是一个结合了大型语言模型（LLMs）与符号计算的神经符号框架。这种创新方法旨在自动化多项式不等式的证明，这一任务传统上需要大量人类专业知识。通过利用机器检查的证明，NSPI不仅增强了这些证明的可扩展性，还弥合了LLM生成的猜想与使用Lean证明助手进行的形式数学验证之间的差距。

**要点：**

- NSPI将LLMs与符号计算结合用于多项式不等式证明。
- 该框架通过机器检查的证明增强了可扩展性。
- 这是自动化形式数学验证的重要一步。

**为什么重要：** 这一发展标志着在自动化复杂数学证明方面的重大飞跃，这可以在各种科学和工程领域中实现更高效的验证过程。

---
Sipply: Sip AI News Simply. https://www.sipply.dev