首页/论文/2012.07106

arXiv 2012.071062020-12math.DG

Geometric Characteristics of Wasserstein Metric on SPD(n)

Yihao Luo, Shiqiang Zhang, Yueqi Cao, Huafei Sun

备注：: 29 pages, 4 figures

英文摘要：

Wasserstein distance, especially among symmetric positive-definite matrices, has broad and deep influences on development of artificial intelligence (AI) and other branches of computer science. A natural idea is to describe the geometry of $SPD\left(n\right)$ as a Riemannian manifold endowed with the Wasserstein metric. In this paper, by involving the fiber bundle, we obtain explicit expressions for some locally geometric quantities, including geodesics, exponential maps, the Riemannian connection, Jacobi fields and curvatures. Furthermore, we discuss the behaviour of geodesics and prove that the manifold is globally geodesic convex with non-negative curvatures but no conjugate pair and cut locus. According to arithmetic estimates, we find curvatures can be controlled by the minimal eigenvalue.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

阅读原文

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

本文内容

英文摘要相关论文

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

继续探索

相关论文

Some geometric calculations on Wasserstein space

We compute the Riemannian connection and curvature for the Wasserstein space of a smooth compact Riemannian manifold.

2007-04-09math.DG

O(n)-invariant Riemannian metrics on SPD matrices

Symmetric Positive Definite (SPD) matrices are ubiquitous in data analysis under the form of covariance matrices or correlation matrices. Several O(n)-invariant Riemannian metrics were defined on the SPD cone, in particular the kernel metrics introduced by Hiai and Petz. The class of kernel metrics interpolates between many classical O(n)-invariant metrics and it satisfies key results of stability and completeness. However, it does not contain all the classical O(n)-invariant metrics. Therefore in this work, we investigate super-classes of kernel metrics and we study which key results remain t…

2021-09-14math.DG

基于广义Bures-Wasserstein几何的对称正定矩阵学习

Learning with symmetric positive definite matrices via generalized Bures-Wasserstein geometry

针对SPD矩阵学习的几何框架需求，本文提出由对称正定矩阵M参数化的广义Bures-Wasserstein（GBW）几何，推导其微分几何性质，并通过实验验证其性能优于原始BW几何，可适配多种机器学习应用。

2023-06-08math.DGmath.FAmath.OC

在SPD流形上通过Cholesky积几何实现快速且稳定的Riemannian度量

Fast and Stable Riemannian Metrics on SPD Manifolds via Cholesky Product Geometry

University of Trento（特伦托大学）；Shanghai Qi Zhi Institute（上海启智研究院）；College of AI, Tsinghua University（清华大学人工智能学院）；Jiangnan University（江南大学）

本文提出两种快速稳定的SPD度量，通过Cholesky分解设计，提升SPD神经网络的计算效率和数值稳定性。

2026-02-08cs.LGmath.DGmath.MG

An intrinsic parallel transport in Wasserstein space

If M is a smooth compact connected Riemannian manifold, let P(M) denote the Wasserstein space of probability measures on M. We describe a geometric construction of parallel transport of some tangent cones along geodesics in P(M). We show that when everything is smooth, the geometric parallel transport agrees with earlier formal calculations.

2017-01-09math.DG

面向SPD矩阵的保持子空间结构的几何统计方法

Geometric statistics with subspace structure preservation for SPD matrices

Nanyang Technological University（南洋理工大学）；University of Cambridge（剑桥大学）；University College London（伦敦大学学院）；University of Tübingen（蒂宾根大学）；KU Leuven（鲁汶大学）

本文提出基于Thompson几何的SPD矩阵处理框架，通过高效计算极端广义特征值保持子空间结构，并研究相应测地空间性质及新型归纳均值。

2024-07-02cs.LGcs.NAmath.DG

更多相关论文收起更多论文

A geometric study of Wasserstein spaces: an addendum on the boundary

We extend the geometric study of the Wasserstein space W(X) of a simply connected, negatively curved metric space X by investigating which pairs of boundary points can be linked by a geodesic, when X is a tree.

2013-05-13math.DGmath.FAmath.MG

Approximate Joint Diagonalization and Geometric Mean of Symmetric Positive Definite Matrices

We explore the connection between two problems that have arisen independently in the signal processing and related fields: the estimation of the geometric mean of a set of symmetric positive definite (SPD) matrices and their approximate joint diagonalization (AJD). Today there is a considerable interest in estimating the geometric mean of a SPD matrix set in the manifold of SPD matrices endowed with the Fisher information metric. The resulting mean has several important invariance properties and has proven very useful in diverse engineering applications such as biomedical and image data proces…

2015-05-26math.DGstat.ML

A geometric study of Wasserstein spaces: isometric rigidity in negative curvature

Given a metric space X, one defines its Wasserstein space W2(X) as a set of sufficiently decaying probability measures on X endowed with a metric defined from optimal transportation. In this article, we continue the geometric study of W2(X) when X is a simply connected, nonpositively curved metric spaces by considering its isometry group. When X is Euclidean, the second named author proved that this isometry group is larger than the isometry group of X. In contrast, we prove here a rigidity result: when X is negatively curved, any isometry of W2(X) comes from an isometry of X.

2019-10-14math.DGmath.MG

Wasserstein--Ebin 度量：不平衡最优输运到黎曼度量空间的几何提升

The Wasserstein--Ebin Metric: A Geometric Lift of Unbalanced Optimal Transport to the space of Riemannian metrics

Florida State University（佛罗里达州立大学）；Universität Wien（维也纳大学）；Université Gustave Eiffel（巴黎-塞纳河大学）

本文通过动态和静态形式化方法，将不平衡最优输运从正密度空间扩展到黎曼度量空间，引入 Wasserstein--Ebin 度量，并证明体积映射定义了到 Wasserstein--Fisher--Rao 度量的黎曼淹没。

2026-05-26math.APmath.DG

A geometric study of Wasserstein spaces: Hadamard spaces

Optimal transport enables one to construct a metric on the set of (sufficiently small at infinity) probability measures on any (not too wild) metric space X, called its Wasserstein space W(X). In this paper we investigate the geometry of W(X) when X is a Hadamard space, by which we mean that $X$ has globally non-positive sectional curvature and is locally compact. Although it is known that -except in the case of the line- W(X) is not non-positively curved, our results show that W(X) have large-scale properties reminiscent of that of X. In particular we define a geodesic boundary for W(X) that …

2013-02-06math.DGmath.MG

On Wasserstein geometry of the space of Gaussian measures

The space of Gaussian measures on a Euclidean space is geodesically convex in the $L^2$-Wasserstein space. This space is a finite dimensional manifold since Gaussian measures are parameterized by means and covariance matrices. By restricting to the space of Gaussian measures inside the $L^2$-Wasserstein space, we manage to provide detailed descriptions of the $L^2$-Wasserstein geometry from a Riemannian geometric viewpoint. We first construct a Riemannian metric which induces the $L^2$-Wasserstein distance. Then we obtain a formula for the sectional curvatures of the space of Gaussian measures…

2009-02-11math.DGmath.PR

正半定锥中基于极端特征值的微分几何

Differential geometry with extreme eigenvalues in the positive semidefinite cone

本文提出基于半定锥Hilbert和Thompson几何中极端广义特征值高效计算的可扩展SPD矩阵分析框架，并定义和证明了一种新型迭代均值的存在性、唯一性及优良性质。

2024-02-08math.DGstat.COstat.ML

Exponential contraction in Wasserstein distance on static and evolving manifolds

In this article, exponential contraction in Wasserstein distance for heat semigroups of diffusion processes on Riemannian manifolds is established under curvature conditions where Ricci curvature is not necessarily required to be non-negative. Compared to the results of Wang (2016), we focus on explicit estimates for the exponential contraction rate. Moreover, we show that our results extend to manifolds evolving under a geometric flow. As application, for the time-inhomogeneous semigroups, we obtain a gradient estimate with an exponential contraction rate under weak curvature conditions, as w…

2020-01-17math.DGmath.PR

SPD流形上约束优化的结构化正则化

Structured Regularization for Constrained Optimization on the SPD Manifold

Harvard University（哈佛大学）

针对SPD流形约束优化中黎曼算法的计算瓶颈，提出基于对称规范函数的结构化正则化器，通过保持凸性或凸差结构，实现用高效无约束方法求解约束问题，数值实验验证了其有效性。

2024-10-12cs.LGmath.DGmath.OC

从主方程中的互易关系出发的概率流形几何计算

Geometric calculations on probability manifolds from reciprocal relations in Master equations

University of South Carolina（南卡罗来纳大学）

本文从主方程互易关系出发，研究概率流形的几何计算，推导了Levi-Civita连接、梯度等算子，并给出了两个具体例子。

2026-02-27math-phmath.COmath.DG