Skip to contents

slicer provides functions for computing sliced Wasserstein distances as well as one-dimensional Wasserstein distances between empirical distributions represented as matrices. These distances can be fed to a Gaussian process regression model, which currently accepts the Gaussian radial basis function kernel as well as the Matérn kernels with smoothness parameters ν=0.5,1.5,2.5\nu = 0.5, 1.5, 2.5.

Installation

You can install the development version of slicer from GitHub with:

# install.packages("pak")
pak::pak("janhove/slicer")

Use

Please refer to the vignette for a brief tutorial.