sinkhorn_pot: Return the dual potentials for the Sinkhorn distance

In approxOT: Approximate and Exact Optimal Transport Methods

Return the dual potentials for the Sinkhorn distance

Description

Return the dual potentials for the Sinkhorn distance

Usage

sinkhorn_pot(
  mass_x,
  mass_y,
  p = 2,
  cost = NULL,
  cost_a = NULL,
  cost_b = NULL,
  ...
)

Arguments

mass_x	The empirical distribution of the first sample
mass_y	The empirical distribution of the second sample
p	The power to raise the cost by
cost	The cost matrix between first and second samples
cost_a	The cost matrix for the first sample
cost_b	The cost matrix for the second sample
...	Additional arguments including epsilon: The fraction of the median cost to use as a penalty niter: Number of iterations to run the Sinkhorn algorithm unbiased: Should the potentials be de-biased TRUE/FALSE

Value

A list with slots "f" and "g", the potentials of the rows and margins, respectively.

approxOT documentation built on May 29, 2024, 3:12 a.m.