Loading…
Alpha Procrustes metrics between positive definite operators: A unifying formulation for the Bures-Wasserstein and Log-Euclidean/Log-Hilbert-Schmidt metrics
This work presents a parametrized family of distances, namely the Alpha Procrustes distances, on the set of symmetric, positive definite (SPD) matrices. The Alpha Procrustes distances provide a unified formulation encompassing both the Bures-Wasserstein and Log-Euclidean distances between SPD matric...
Saved in:
Published in: | Linear algebra and its applications 2022-03, Vol.636, p.25-68 |
---|---|
Main Author: | |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites Items that cite this one |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
cited_by | cdi_FETCH-LOGICAL-c391t-af7d1f6119272ffc6bec66cbcdc81429ae8886e4ab424189d22c8d088cb4dff93 |
---|---|
cites | cdi_FETCH-LOGICAL-c391t-af7d1f6119272ffc6bec66cbcdc81429ae8886e4ab424189d22c8d088cb4dff93 |
container_end_page | 68 |
container_issue | |
container_start_page | 25 |
container_title | Linear algebra and its applications |
container_volume | 636 |
creator | Minh, Hà Quang |
description | This work presents a parametrized family of distances, namely the Alpha Procrustes distances, on the set of symmetric, positive definite (SPD) matrices. The Alpha Procrustes distances provide a unified formulation encompassing both the Bures-Wasserstein and Log-Euclidean distances between SPD matrices. We show that the Alpha Procrustes distances are the Riemannian distances corresponding to a family of Riemannian metrics on the manifold of SPD matrices, which encompass both the Log-Euclidean and Wasserstein Riemannian metrics. This formulation is then generalized to the set of positive definite Hilbert-Schmidt operators on a Hilbert space, unifying the infinite-dimensional Bures-Wasserstein and Log-Hilbert-Schmidt distances. In the setting of reproducing kernel Hilbert spaces (RKHS) covariance operators, we obtain closed form formulas for all the distances via the corresponding kernel Gram matrices. From a statistical viewpoint, the Alpha Procrustes distances give rise to a parametrized family of distances between Gaussian measures on Euclidean space, in the finite-dimensional case, and separable Hilbert spaces, in the infinite-dimensional case, encompassing the 2-Wasserstein distance, with closed form formulas via Gram matrices in the RKHS setting. The presented formulations are new both in the finite and infinite-dimensional settings. |
doi_str_mv | 10.1016/j.laa.2021.11.011 |
format | article |
fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_journals_2652655908</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0024379521004110</els_id><sourcerecordid>2652655908</sourcerecordid><originalsourceid>FETCH-LOGICAL-c391t-af7d1f6119272ffc6bec66cbcdc81429ae8886e4ab424189d22c8d088cb4dff93</originalsourceid><addsrcrecordid>eNp9kcGKFDEQhoMoOK4-gLeA5-5NZXq603oal9UVBnZhVzyGdFLZqaEnaZP0yr6LD2sPo1ehoKrg_6t--Bh7D6IGAe3loR6NqaWQUAPUAuAFW4Hq1hWoTfuSrYSQTbXu-s1r9ibngxCi6YRcsd_bcdobfpeiTXMumPkRSyKb-YDlF2LgU8xU6Am5Q0-BCvI4YTIlpvyRb_kcyD9TeOQ-puM8mkIxnGZe9sg_zwlz9cPkjGk5ToGb4PguPlbXsx3JoQmXp-2GxgFTqe7t_kiu_Mvwlr3yZsz47m-_YN-_XD9c3VS726_frra7yq57KJXxnQPfAvSyk97bdkDbtnawzipoZG9QKdViY4ZGNqB6J6VVTihlh8Z5368v2Ifz3SnFnzPmog9xTmF5qWW7WWrTC7Wo4KyyKeac0Osp0dGkZw1CnyDog14g6BMEDaAXCIvn09mDS_wnwqSzJQwWHSW0RbtI_3H_Ac6ak1g</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2652655908</pqid></control><display><type>article</type><title>Alpha Procrustes metrics between positive definite operators: A unifying formulation for the Bures-Wasserstein and Log-Euclidean/Log-Hilbert-Schmidt metrics</title><source>ScienceDirect Journals</source><creator>Minh, Hà Quang</creator><creatorcontrib>Minh, Hà Quang</creatorcontrib><description>This work presents a parametrized family of distances, namely the Alpha Procrustes distances, on the set of symmetric, positive definite (SPD) matrices. The Alpha Procrustes distances provide a unified formulation encompassing both the Bures-Wasserstein and Log-Euclidean distances between SPD matrices. We show that the Alpha Procrustes distances are the Riemannian distances corresponding to a family of Riemannian metrics on the manifold of SPD matrices, which encompass both the Log-Euclidean and Wasserstein Riemannian metrics. This formulation is then generalized to the set of positive definite Hilbert-Schmidt operators on a Hilbert space, unifying the infinite-dimensional Bures-Wasserstein and Log-Hilbert-Schmidt distances. In the setting of reproducing kernel Hilbert spaces (RKHS) covariance operators, we obtain closed form formulas for all the distances via the corresponding kernel Gram matrices. From a statistical viewpoint, the Alpha Procrustes distances give rise to a parametrized family of distances between Gaussian measures on Euclidean space, in the finite-dimensional case, and separable Hilbert spaces, in the infinite-dimensional case, encompassing the 2-Wasserstein distance, with closed form formulas via Gram matrices in the RKHS setting. The presented formulations are new both in the finite and infinite-dimensional settings.</description><identifier>ISSN: 0024-3795</identifier><identifier>EISSN: 1873-1856</identifier><identifier>DOI: 10.1016/j.laa.2021.11.011</identifier><language>eng</language><publisher>Amsterdam: Elsevier Inc</publisher><subject>Closed form solutions ; Covariance operators ; Euclidean geometry ; Euclidean space ; Exact solutions ; Formulations ; Gaussian measures ; Hilbert space ; Kernels ; Linear algebra ; Log-Euclidean distance ; Log-Hilbert-Schmidt distance ; Mathematical analysis ; Matrices (mathematics) ; Operators (mathematics) ; Parameterization ; Positive definite matrices and operators ; Procrustes distance ; Reproducing kernel Hilbert spaces ; Wasserstein distance</subject><ispartof>Linear algebra and its applications, 2022-03, Vol.636, p.25-68</ispartof><rights>2021 Elsevier Inc.</rights><rights>Copyright American Elsevier Company, Inc. Mar 1, 2022</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c391t-af7d1f6119272ffc6bec66cbcdc81429ae8886e4ab424189d22c8d088cb4dff93</citedby><cites>FETCH-LOGICAL-c391t-af7d1f6119272ffc6bec66cbcdc81429ae8886e4ab424189d22c8d088cb4dff93</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids></links><search><creatorcontrib>Minh, Hà Quang</creatorcontrib><title>Alpha Procrustes metrics between positive definite operators: A unifying formulation for the Bures-Wasserstein and Log-Euclidean/Log-Hilbert-Schmidt metrics</title><title>Linear algebra and its applications</title><description>This work presents a parametrized family of distances, namely the Alpha Procrustes distances, on the set of symmetric, positive definite (SPD) matrices. The Alpha Procrustes distances provide a unified formulation encompassing both the Bures-Wasserstein and Log-Euclidean distances between SPD matrices. We show that the Alpha Procrustes distances are the Riemannian distances corresponding to a family of Riemannian metrics on the manifold of SPD matrices, which encompass both the Log-Euclidean and Wasserstein Riemannian metrics. This formulation is then generalized to the set of positive definite Hilbert-Schmidt operators on a Hilbert space, unifying the infinite-dimensional Bures-Wasserstein and Log-Hilbert-Schmidt distances. In the setting of reproducing kernel Hilbert spaces (RKHS) covariance operators, we obtain closed form formulas for all the distances via the corresponding kernel Gram matrices. From a statistical viewpoint, the Alpha Procrustes distances give rise to a parametrized family of distances between Gaussian measures on Euclidean space, in the finite-dimensional case, and separable Hilbert spaces, in the infinite-dimensional case, encompassing the 2-Wasserstein distance, with closed form formulas via Gram matrices in the RKHS setting. The presented formulations are new both in the finite and infinite-dimensional settings.</description><subject>Closed form solutions</subject><subject>Covariance operators</subject><subject>Euclidean geometry</subject><subject>Euclidean space</subject><subject>Exact solutions</subject><subject>Formulations</subject><subject>Gaussian measures</subject><subject>Hilbert space</subject><subject>Kernels</subject><subject>Linear algebra</subject><subject>Log-Euclidean distance</subject><subject>Log-Hilbert-Schmidt distance</subject><subject>Mathematical analysis</subject><subject>Matrices (mathematics)</subject><subject>Operators (mathematics)</subject><subject>Parameterization</subject><subject>Positive definite matrices and operators</subject><subject>Procrustes distance</subject><subject>Reproducing kernel Hilbert spaces</subject><subject>Wasserstein distance</subject><issn>0024-3795</issn><issn>1873-1856</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><recordid>eNp9kcGKFDEQhoMoOK4-gLeA5-5NZXq603oal9UVBnZhVzyGdFLZqaEnaZP0yr6LD2sPo1ehoKrg_6t--Bh7D6IGAe3loR6NqaWQUAPUAuAFW4Hq1hWoTfuSrYSQTbXu-s1r9ibngxCi6YRcsd_bcdobfpeiTXMumPkRSyKb-YDlF2LgU8xU6Am5Q0-BCvI4YTIlpvyRb_kcyD9TeOQ-puM8mkIxnGZe9sg_zwlz9cPkjGk5ToGb4PguPlbXsx3JoQmXp-2GxgFTqe7t_kiu_Mvwlr3yZsz47m-_YN-_XD9c3VS726_frra7yq57KJXxnQPfAvSyk97bdkDbtnawzipoZG9QKdViY4ZGNqB6J6VVTihlh8Z5368v2Ifz3SnFnzPmog9xTmF5qWW7WWrTC7Wo4KyyKeac0Osp0dGkZw1CnyDog14g6BMEDaAXCIvn09mDS_wnwqSzJQwWHSW0RbtI_3H_Ac6ak1g</recordid><startdate>20220301</startdate><enddate>20220301</enddate><creator>Minh, Hà Quang</creator><general>Elsevier Inc</general><general>American Elsevier Company, Inc</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>20220301</creationdate><title>Alpha Procrustes metrics between positive definite operators: A unifying formulation for the Bures-Wasserstein and Log-Euclidean/Log-Hilbert-Schmidt metrics</title><author>Minh, Hà Quang</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c391t-af7d1f6119272ffc6bec66cbcdc81429ae8886e4ab424189d22c8d088cb4dff93</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Closed form solutions</topic><topic>Covariance operators</topic><topic>Euclidean geometry</topic><topic>Euclidean space</topic><topic>Exact solutions</topic><topic>Formulations</topic><topic>Gaussian measures</topic><topic>Hilbert space</topic><topic>Kernels</topic><topic>Linear algebra</topic><topic>Log-Euclidean distance</topic><topic>Log-Hilbert-Schmidt distance</topic><topic>Mathematical analysis</topic><topic>Matrices (mathematics)</topic><topic>Operators (mathematics)</topic><topic>Parameterization</topic><topic>Positive definite matrices and operators</topic><topic>Procrustes distance</topic><topic>Reproducing kernel Hilbert spaces</topic><topic>Wasserstein distance</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Minh, Hà Quang</creatorcontrib><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>Linear algebra and its applications</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Minh, Hà Quang</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Alpha Procrustes metrics between positive definite operators: A unifying formulation for the Bures-Wasserstein and Log-Euclidean/Log-Hilbert-Schmidt metrics</atitle><jtitle>Linear algebra and its applications</jtitle><date>2022-03-01</date><risdate>2022</risdate><volume>636</volume><spage>25</spage><epage>68</epage><pages>25-68</pages><issn>0024-3795</issn><eissn>1873-1856</eissn><abstract>This work presents a parametrized family of distances, namely the Alpha Procrustes distances, on the set of symmetric, positive definite (SPD) matrices. The Alpha Procrustes distances provide a unified formulation encompassing both the Bures-Wasserstein and Log-Euclidean distances between SPD matrices. We show that the Alpha Procrustes distances are the Riemannian distances corresponding to a family of Riemannian metrics on the manifold of SPD matrices, which encompass both the Log-Euclidean and Wasserstein Riemannian metrics. This formulation is then generalized to the set of positive definite Hilbert-Schmidt operators on a Hilbert space, unifying the infinite-dimensional Bures-Wasserstein and Log-Hilbert-Schmidt distances. In the setting of reproducing kernel Hilbert spaces (RKHS) covariance operators, we obtain closed form formulas for all the distances via the corresponding kernel Gram matrices. From a statistical viewpoint, the Alpha Procrustes distances give rise to a parametrized family of distances between Gaussian measures on Euclidean space, in the finite-dimensional case, and separable Hilbert spaces, in the infinite-dimensional case, encompassing the 2-Wasserstein distance, with closed form formulas via Gram matrices in the RKHS setting. The presented formulations are new both in the finite and infinite-dimensional settings.</abstract><cop>Amsterdam</cop><pub>Elsevier Inc</pub><doi>10.1016/j.laa.2021.11.011</doi><tpages>44</tpages></addata></record> |
fulltext | fulltext |
identifier | ISSN: 0024-3795 |
ispartof | Linear algebra and its applications, 2022-03, Vol.636, p.25-68 |
issn | 0024-3795 1873-1856 |
language | eng |
recordid | cdi_proquest_journals_2652655908 |
source | ScienceDirect Journals |
subjects | Closed form solutions Covariance operators Euclidean geometry Euclidean space Exact solutions Formulations Gaussian measures Hilbert space Kernels Linear algebra Log-Euclidean distance Log-Hilbert-Schmidt distance Mathematical analysis Matrices (mathematics) Operators (mathematics) Parameterization Positive definite matrices and operators Procrustes distance Reproducing kernel Hilbert spaces Wasserstein distance |
title | Alpha Procrustes metrics between positive definite operators: A unifying formulation for the Bures-Wasserstein and Log-Euclidean/Log-Hilbert-Schmidt metrics |
url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T06%3A16%3A34IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Alpha%20Procrustes%20metrics%20between%20positive%20definite%20operators:%20A%20unifying%20formulation%20for%20the%20Bures-Wasserstein%20and%20Log-Euclidean/Log-Hilbert-Schmidt%20metrics&rft.jtitle=Linear%20algebra%20and%20its%20applications&rft.au=Minh,%20H%C3%A0%20Quang&rft.date=2022-03-01&rft.volume=636&rft.spage=25&rft.epage=68&rft.pages=25-68&rft.issn=0024-3795&rft.eissn=1873-1856&rft_id=info:doi/10.1016/j.laa.2021.11.011&rft_dat=%3Cproquest_cross%3E2652655908%3C/proquest_cross%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c391t-af7d1f6119272ffc6bec66cbcdc81429ae8886e4ab424189d22c8d088cb4dff93%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2652655908&rft_id=info:pmid/&rfr_iscdi=true |