Compute log second marginal and transported feature means (dense)

Given translation-invariant dual potentials (fbar,gbar), compute the log of the second marginal (column sums) of the entropic UOT coupling and the feature barycentric projection means per target node.

Usage

uot_kl_logpi2_meanF_dense_cpp(cost, alpha, beta, fbar, gbar, epsilon, F = NULL)

Arguments

cost

Dense cost matrix (n x m).

alpha

Source masses (length n).

beta

Target masses (length m).

fbar

Translation-invariant source potential (length n).

gbar

Translation-invariant target potential (length m).

epsilon

Entropic regularization parameter (> 0).

F

Optional source features (n x D). If provided, returns mean features.

Value

A list with `log_pi2` (length m) and optionally `mean_F` (m x D).