ott.problems.linear.potentials.EntropicPotentials

ott.problems.linear.potentials.EntropicPotentials#

class ott.problems.linear.potentials.EntropicPotentials(f_xy, g_xy, prob, f_xx=None, g_yy=None)[source]#

Dual potential functions from finite samples [Pooladian and Niles-Weed, 2021].

Parameters:
  • f_xy (Array) – The first dual potential vector of shape [n,].

  • g_xy (Array) – The second dual potential vector of shape [m,].

  • prob (LinearProblem) – Linear problem with PointCloud geometry that was used to compute the dual potentials using, e.g., Sinkhorn.

  • f_xx (Optional[Array]) – The first dual potential vector of shape [n,] used for debiasing [Pooladian et al., 2022].

  • g_yy (Optional[Array]) – The second dual potential vector of shape [m,] used for debiasing.

Methods

distance(src, tgt)

Evaluate Wasserstein distance between samples using dual potentials.

plot_ot_map(source, target[, samples, ...])

Plot data and learned optimal transport map.

plot_potential([forward, quantile, ...])

Plot the potential.

transport(vec[, forward])

Transport vec according to Gangbo-McCann Brenier [Brenier, 1991].

Attributes

epsilon

Entropy regularizer.

f

The first dual potential function.

g

The second dual potential function.

is_debiased

Whether the entropic map is debiased.