LOMAR (version 0.5.0)

get_kernel_matrix: get_kernel_matrix

Description

Compute kernel/distance matrix between persistence diagrams.

Usage

get_kernel_matrix(
  Diag = NULL,
  method = c("sWd", "pssk"),
  dimensions = NULL,
  return.dist = FALSE,
  M = NULL,
  sigma = NULL,
  ncpu = 1,
  cluster.type = "PSOCK"
)

Value

a matrix

Arguments

Diag

list of persistence diagrams as n x 3 matrices

method

which kernel or distance to compute. One of sWd (for sliced Wasserstein kernel) or pssk (for the persistence scale-space kernel)

dimensions

vector of the dimensions of the topological features to consider, if NULL (default) use all available dimensions

return.dist

logical (default: FALSE) for method sWd, whether to return the sliced Wasserstein distance matrix instead of the kernel.

M

number of slices for the sliced Wasserstein kernel

sigma

kernel bandwidth

ncpu

number of parallel threads to use for computation

cluster.type

type of multicore cluster to use, either PSOCK (default) or FORK

Examples

PS <- list(data.frame(x = c(2.4,-6.9,4.6,-0.7,-3.3,-4.9,-3.5,-3.5,4.2,-7),
                      y = c(5.7,1.9,4.8,3.4,-3,-2.1,7.2,1.8,6.1,-1.6),
                      z = c(2.7,-0.1,-0.7,-0.6,0.4,-1.5,-0.6,-0.9,2.2,0.7)),
           data.frame(x = c(0,0,3.1,-5.6,-5,-7.4,-0.7,-7.7,-6.7,4,4.2,0.2,5.8,3.9,3.9),
                      y = c(6.3,-6.1,-3.5,4.6,-4.1,0.3,8.8,-2.3,2.9,3.7,-1.4,-3.9,5.5,-1.2,-6.7),
                      z = c(-1.5,1.7,-0.4,-1.4,1.8,1.7,-0.9,-1.8,-0.5,1.7,1.3,0.5,-1.4,1.6,-0.1)),
           data.frame(x = c(-9.8,-5.2,12.5,2.5,4.5,1.3,-0.2,0.4,9.3,-1.4,0.5,-1.1,-7.7),
                      y = c(-4.2,1.5,-0.5,12,-3,-7.2,10.9,6.7,-1.3,10,6.7,-6.2,2.9),
                      z = c(3.4,-3.8,-1.4,1.8,3.5,2.5,2.6,-4.8,-3.8,3.9,4.1,-3.6,-4)))
Dgs <- get_persistence_diagrams(point.sets = PS, maxdimension = 1, maxscale = 5, ncpu = 1)
K <- get_kernel_matrix(Diag = Dgs, method = 'sWd', dimensions = c(0,1), M = 10, sigma = 5)