transport_plan_multimarg {approxOT}

Multimarginal optimal transport plans

Description

Multimarginal optimal transport plans

Usage

transport_plan_multimarg(
  ...,
  p = 2,
  ground_p = 2,
  observation.orientation = c("rowwise", "colwise"),
  method = c("hilbert", "univariate", "sliced"),
  nsim = 1000
)

Arguments

...

Either data matrices as separate arguments or a list of data matrices. Arguments after the data must be specified by name.

p

The power of the Wasserstein distance to use

ground_p

The power of the Euclidean distance to use

observation.orientation

Are observations by rows or columns

method

One of "hilbert", "univariate", or "sliced"

nsim

Number of simulations to use for the sliced method

Value

transport plan

Examples

set.seed(23423)
n <- 100
d <- 10
p <- ground_p <- 2 #euclidean cost, p = 2
x <- matrix(stats::rnorm((n + 11)*d), n + 11 , d)
y <- matrix(stats::rnorm(n*d), n, d)
z <- matrix(stats::rnorm((n +455)*d), n +455, d)

# make data a list
data <- list(x,y,z)

tplan <- transport_plan_multimarg(data, p = p, ground_p = ground_p,
observation.orientation = "rowwise", method = "hilbert")

#' #transpose data works too
datat <- lapply(data, t)

tplan2 <- transport_plan_multimarg(datat, p = p, ground_p = ground_p,
observation.orientation = "colwise",method = "hilbert")