OptimalTransportTools

applies a ↦ b = Ka

applies a ↦ b = Ka

getcostmatrix(c)

c: separated cost matrix as a dx(nxn) vector of matrices

returns cost matrix C in non-separated form as a NxN matrix

TODO: different n along every dimension, d > 2

getcostmatrix_separated(n, d; a=0, b=0) n: number of discretization points along one dimension a: left interval bound b: right interval bound

returns cost matrix C in separated form as a dx(nxn) vector of matrices

TODO: non-cube-domains, different n along every dimension

!!! Averaged updates not implemented here

Convenience function for no given initial potentials

This maps (f, c, log_α) ↦ g = - ε log.(∑ₓ α .* exp ε⁻¹ ( f - C(.,y) ))