Anisotropic mark correlation function, direction vector formulation

Estimate the anisotropic mark correlation function (2d and 3d), as defined in Stoyan 1991, f. 5.4. (without the additional 'a'-marks)

Usage

markcorr_anisotropic(
  x,
  marks = NULL,
  r,
  f = function(a, b) a * b,
  directions,
  bw,
  adjust = c(1, 1),
  correction = "translation",
  bootsize = 1e+05,
  divisor = "d",
  normaliser = NULL
)

Arguments

x: pp, list with $x~coordinates $bbox~bounding box
marks: if x is not marked (x$marks is empty), use these marks
r: Evaluate at these lengths.
f: test function of the form function(m1, m2) ..., returning a vector of length(m1)
directions: Matrix of directions, in unit vectors, one direction per row. Default: along axes.
bw: Bandwidths of epanechnicov kernels. Vector of two values, one for ranges and one for angles.
adjust: If bw not given, use bw=adjust[1]*0.15/lambda^(1/dim) for range and bw=adjust[2]*0.15*pi for angle
correction: "none" or "translation". Translation only for rectangular box.
bootsize: bootstrap size for estimating the normaliser if not given.
divisor: either "d" or "r". Divide by dist(i,j) ("d") instead of r ("r")?
normaliser: normalising constant under independent marking. If NULL, estimated with bootstrap.

Details

Default bandwidth: bw=adjust*0.15/lambda^(1/dim) for range and bw=adjust*0.15*pi for angle.