Curvature of the Manifold of Fixed-Rank Positive-Semidefinite Matrices Endowed with the Bures-Wasserstein Metric

被引:9
|
作者
Massart, Estelle [1 ]
Hendrickx, Julien M. [1 ,2 ]
Absil, P-A [1 ]
机构
[1] UCLouvain, ICTEAM, Louvain La Neuve, Belgium
[2] Boston Univ, Boston, MA 02215 USA
来源
关键词
OPTIMIZATION; GEOMETRY;
D O I
10.1007/978-3-030-26980-7_77
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We consider the manifold of rank-p positive-semidefinite matrices of size n, seen as a quotient of the set of full-rank n-by-p matrices by the orthogonal group in dimension p. The resulting distance coincides with theWasserstein distance between centered degenerate Gaussian distributions. We obtain expressions for the Riemannian curvature tensor and the sectional curvature of the manifold. We also provide tangent vectors spanning planes associated with the extreme values of the sectional curvature.
引用
收藏
页码:739 / 748
页数:10
相关论文
共 13 条
  • [1] QUOTIENT GEOMETRY WITH SIMPLE GEODESICS FOR THE MANIFOLD OF FIXED-RANK POSITIVE-SEMIDEFINITE MATRICES
    Massart, Estelle
    Absil, P-A
    [J]. SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 2020, 41 (01) : 171 - 198
  • [2] Fixed-Rank Approximation of a Positive-Semidefinite Matrix from Streaming Data
    Tropp, Joel A.
    Yurtsever, Alp
    Udell, Madeleine
    Cevher, Volkan
    [J]. ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 30 (NIPS 2017), 2017, 30
  • [3] On the Bures-Wasserstein distance between positive definite matrices
    Bhatia, Rajendra
    Jain, Tanvi
    Lim, Yongdo
    [J]. EXPOSITIONES MATHEMATICAE, 2019, 37 (02) : 165 - 191
  • [4] Regression on Fixed-Rank Positive Semidefinite Matrices: A Riemannian Approach
    Meyer, Gilles
    Bonnabel, Silvere
    Sepulchre, Rodolphe
    [J]. JOURNAL OF MACHINE LEARNING RESEARCH, 2011, 12 : 593 - 625
  • [5] Regression on fixed-rank positive semidefinite matrices: A Riemannian approach
    Meyer, Gilles
    Bonnabel, Silvère
    Sepulchre, Rodolphe
    [J]. Journal of Machine Learning Research, 2011, 12 : 593 - 625
  • [6] Two Newton methods on the manifold of fixed-rank matrices endowed with Riemannian quotient geometries
    Absil, P-A
    Amodei, Luca
    Meyer, Gilles
    [J]. COMPUTATIONAL STATISTICS, 2014, 29 (3-4) : 569 - 590
  • [7] On Riemannian Optimization over Positive Definite Matrices with the Bures-Wasserstein Geometry
    Han, Andi
    Mishra, Bamdev
    Jawanpuria, Pratik
    Gao, Junbin
    [J]. ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 34 (NEURIPS 2021), 2021, 34
  • [8] Two Newton methods on the manifold of fixed-rank matrices endowed with Riemannian quotient geometries
    P.-A. Absil
    Luca Amodei
    Gilles Meyer
    [J]. Computational Statistics, 2014, 29 : 569 - 590
  • [9] Fixed-Rank Supervised Metric Learning on Riemannian Manifold
    Mu, Yadong
    [J]. THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2016, : 1941 - 1947
  • [10] RIEMANNIAN METRIC AND GEOMETRIC MEAN FOR POSITIVE SEMIDEFINITE MATRICES OF FIXED RANK
    Bonnabel, Silvere
    Sepulchre, Rodolphe
    [J]. SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 2009, 31 (03) : 1055 - 1070