Accelerating the Sinkhorn Algorithm for Sparse Multi-Marginal Optimal Transport via Fast Fourier Transforms

被引:7
作者
Ba, Fatima Antarou [1 ]
Quellmalz, Michael [1 ]
机构
[1] Tech Univ Berlin, Inst Math, Str 17 Juni 136, D-10623 Berlin, Germany
关键词
multi-marginal optimal transport; Sinkhorn algorithm; fast Fourier transform; image processing; optimal transport; Wasserstein barycenter; SUMMATION;
D O I
10.3390/a15090311
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 numerical solution of the discrete multi-marginal optimal transport (MOT) by means of the Sinkhorn algorithm. In general, the Sinkhorn algorithm suffers from the curse of dimensionality with respect to the number of marginals. If the MOT cost function decouples according to a tree or circle, its complexity is linear in the number of marginal measures. In this case, we speed up the convolution with the radial kernel required in the Sinkhorn algorithm via non-uniform fast Fourier methods. Each step of the proposed accelerated Sinkhorn algorithm with a tree-structured cost function has a complexity of O (KN) instead of the classical O (KN2) for straightforward matrix- vector operations, where K is the number of marginals and each marginal measure is supported on, at most, N points. In the case of a circle-structured cost function, the complexity improves from O (KN3) to O (KN2). This is confirmed through numerical experiments.
引用
收藏
页数:19
相关论文
共 58 条
  • [11] Generalized incompressible flows, multi-marginal transport and Sinkhorn algorithm
    Benamou, Jean-David
    Carlier, Guillaume
    Nenna, Luca
    [J]. NUMERISCHE MATHEMATIK, 2019, 142 (01) : 33 - 54
  • [12] Benamou JD, 2016, SCI COMPUT, P578, DOI 10.1007/978-3-319-41589-5_17
  • [13] ITERATIVE BREGMAN PROJECTIONS FOR REGULARIZED TRANSPORTATION PROBLEMS
    Benamou, Jean-David
    Carlier, Guillaume
    Cuturi, Marco
    Nenna, Luca
    Peyre, Gabriel
    [J]. SIAM JOURNAL ON SCIENTIFIC COMPUTING, 2015, 37 (02) : A1111 - A1138
  • [14] ON THE FAST FOURIER-TRANSFORM OF FUNCTIONS WITH SINGULARITIES
    BEYLKIN, G
    [J]. APPLIED AND COMPUTATIONAL HARMONIC ANALYSIS, 1995, 2 (04) : 363 - 381
  • [15] Blondel Mathieu, 2018, INT C ART INT STAT, V84, P880
  • [16] Wasserstein Barycentric Coordinates: Histogram Regression Using Optimal Transport
    Bonneel, Nicolas
    Peyre, Gabriel
    Cuturi, Marco
    [J]. ACM TRANSACTIONS ON GRAPHICS, 2016, 35 (04):
  • [17] Brenier Y, 1999, COMMUN PUR APPL MATH, V52, P411, DOI 10.1002/(SICI)1097-0312(199904)52:4<411::AID-CPA1>3.0.CO
  • [18] 2-3
  • [19] THE DUAL LEAST ACTION PROBLEM FOR AN IDEAL, INCOMPRESSIBLE FLUID
    BRENIER, Y
    [J]. ARCHIVE FOR RATIONAL MECHANICS AND ANALYSIS, 1993, 122 (04) : 323 - 351
  • [20] NUMERICAL METHODS FOR MATCHING FOR TEAMS AND WASSERSTEIN BARYCENTERS
    Carlier, Guillaume
    Oberman, Adam
    Oudet, Edouard
    [J]. ESAIM-MATHEMATICAL MODELLING AND NUMERICAL ANALYSIS-MODELISATION MATHEMATIQUE ET ANALYSE NUMERIQUE, 2015, 49 (06): : 1621 - 1642