arXiv:2511.22882cs.LGmath.PR2025-11

提出边界商构造新流形密度模型,可高效建模对称结构数据。

Normalizing Flows on Quotient Manifolds via Boundary Quotients

  • 基于边界商构造流形上的归一化流,支持对称性建模。
  • 在环面与透镜空间上实验,计算成本更低且性能强。
  • 适合需要对称感知生成建模的研究者使用。

我们引入边界商并提出一种学习边界商简单域上流形密度的框架。该框架可用于构建离散群 $G$ 作用于流形 $N$ 时的商流形 $N/G$ 上的归一化流。我们在亏格为 $g$ 的曲面 $Σ_g$ 上实例化该构造。当 $G$ 有限时,该方法适用于对称性感知学习,我们在 3-球的循环商上进行了验证。在透镜空间上的实验表明,简单的预商 RealNVP 模型即可取得优异结果,且评估开销显著降低。

原文摘要 · Abstract (English)

We introduce boundary quotients and present a framework for learning densities on manifolds that arise as boundary quotients of simpler domains. We show that this framework can be used to construct normalizing flows on quotient manifolds $N/G$, where a discrete group $G$ acts on $N$. We instantiate this construction for genus-$g$ surfaces $Σ_g$. When $G$ is finite, we show applicability to symmetry aware learning; we demonstrate this on cyclic quotients of the 3-sphere. Experiments on lens spaces show that simple pre-quotient RealNVP models can achieve strong results while being substantially cheaper to evaluate.

归一化流流形学习对称性建模

Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。