Skip to main content
QUICK REVIEW

[论文解读] Flow-based Extremal Mathematical Structure Discovery

Gergely Bérczi, Baran Hashemi|arXiv (Cornell University)|Jan 25, 2026
Machine Learning in Materials Science被引用 0
一句话总结

FlowBoost 是一个闭环、基于流的生成框架,通过整合条件流匹配、几何感知采样和以奖励为导向的微调,学习发现罕见的极端几何配置,在显著减少计算资源且不依赖大语言模型的情况下实现具有竞争力的结果。

ABSTRACT

The discovery of extremal structures in mathematics requires navigating vast and nonconvex landscapes where analytical methods offer little guidance and brute-force search becomes intractable. We introduce FlowBoost, a closed-loop generative framework that learns to discover rare and extremal geometric structures by combining three components: (i) a geometry-aware conditional flow-matching model that learns to sample high-quality configurations, (ii) reward-guided policy optimization with action exploration that directly optimizes the generation process toward the objective while maintaining diversity, and (iii) stochastic local search for both training-data generation and final refinement. Unlike prior open-loop approaches, such as PatternBoost that retrains on filtered discrete samples, or AlphaEvolve which relies on frozen Large Language Models (LLMs) as evolutionary mutation operators, FlowBoost enforces geometric feasibility during sampling, and propagates reward signal directly into the generative model, closing the optimization loop and requiring much smaller training sets and shorter training times, and reducing the required outer-loop iterations by orders of magnitude, while eliminating dependence on LLMs. We demonstrate the framework on four geometric optimization problems: sphere packing in hypercubes, circle packing maximizing sum of radii, the Heilbronn triangle problem, and star discrepancy minimization. In several cases, FlowBoost discovers configurations that match or exceed the best known results. For circle packings, we improve the best known lower bounds, surpassing the LLM-based system AlphaEvolve while using substantially fewer computational resources.

研究动机与目标

  • 动机化并形式化在高维配置空间中发现极值几何结构的问题。
  • 将 FlowBoost 引入为一个闭环 SBO 框架,结合连续生成、几何感知采样与奖励驱动优化。
  • 在球面装填、圆形堆放、海尔布伦问题和星形离散度上展示 FlowBoost,以获得具有竞争力或更优的结果。
  • 展示领域特定的归纳偏置和奖励驱动学习在较低计算成本下能够匹配或超越基于LLM的系统。

提出的方法

  • 采用条件流匹配来学习一个随时间变化的向量场,将简单先验运输到高质量配置。
  • 通过将流积分与对约束流形的投影交错进行,纳入几何感知采样以保持可行性。
  • 在训练过程中整合几何惩罚和重叠能量以强制执行约束(软惩罚与投影)。
  • 使用带教师-学生设置的奖励驱动微调(RG-CFM)和基于重要性加权的更新,将采样偏向高奖励配置,同时避免塌缩。
  • 采用基于 GAS 的推断过程,具有自适应墙、Gauss–Newton 投影、近端松弛和端部细化,以确保可行样本。
  • 定义面向问题的奖励(例如用于打包的有效半径),并使用带一致性正则化的加权流匹配损失,在提高性能的同时保持多样性。
Figure 2 . Comparison of the minimum-excess metric for two local search heuristics (common circle counts). The physics-push heuristic yields consistently higher minimum-excess values than SRP/SRS, indicating worse configurations across the tested regime.
Figure 2 . Comparison of the minimum-excess metric for two local search heuristics (common circle counts). The physics-push heuristic yields consistently higher minimum-excess values than SRP/SRS, indicating worse configurations across the tested regime.

实验结果

研究问题

  • RQ1在连续空间中,是否可以利用具有几何感知采样的基于流的生成模型有效地发现极值几何配置?
  • RQ2在经典极值几何问题上,闭环奖励驱动的 SBO 框架是否优于开环生成方法和基于LLM的方法?
  • RQ3几何约束和归纳偏置如何影响样本的可行性及对罕见配置的发现?
  • RQ4FlowBoost 在多大程度上可以改进球面堆放、圆形堆放、海尔布伦问题和星形离散度问题的已知构造?

主要发现

  • FlowBoost 在若干几何优化问题上发现的配置与到目前为止的最佳结果相匹配或超越。
  • 对于圆形打包,FlowBoost 改善了已知的最佳下界,并在更低的计算资源下超越基于LLM 的 AlphaEvolve。
  • 在海尔布伦问题实例中,FlowBoost 将最小三角形面积提升至接近已知的数值值。
  • 在维度12的球面打包中,FlowBoost 找到了比某些经典启发式更密集的配置。
  • FlowBoost 被提出作为首个将基于流的生成模型与强化学习系统应用于极值数学的系统性尝试,在减少计算量且不使用 LLM 的情况下取得了优异结果。
(a) $N=55$ : training vs. RG-CFM (pushed).
(a) $N=55$ : training vs. RG-CFM (pushed).

更好的研究,从现在开始

从阅读论文到最终审阅,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。