optimal_transport: solve the Optimal transport problem

In jagterberg/optimalTransportInvariance: Nonparametric Testing for Random Graphs

Description

Optimally transport according to wasserstein cost C_ij = d(QX_i, Y_j)^2. this is the usual optimal transport problem except with an allowance of a parameter Q for an orthogonal matrix. Give two matrices of dimensions n x d and m x y respectively

Usage

optimal_transport(X, Y, Q = NULL, lambda = 0.1, eps = 0.01)

Arguments

X	an n x d matrix of points where each row is a point
Y	similar, except possibly a different number of points
Q	An orthogonal matrix
lambda	the parameter to penalize in sinkhorn divergence
eps	the tolerance

Value

P, the n x m matrix of assignments

jagterberg/optimalTransportInvariance documentation built on Feb. 3, 2022, 7:24 p.m.