Simulation of Poisson cluster process on linear network

Question

On pp. 73-74 in "“Stationary” point processes are uncommon on linear networks"* by Adrian Baddeley, Gopolan Nair, Suman Rakshit, and Greg McSwiggan, the authors introduce a Poisson cluster process on a linear network and subsequently demonstrate a simulation.

Unfortunately, I cannot find related code. Does {spatstat} provide an algorithm to simulate such a process?

On pp. 24-25 in "Analysing point patterns on networks — A review"* by the aforementioned authors and Tilman M. Davies one can read:

"Point process models which exhibit clustering (positive association between points), such as Poisson cluster processes and Cox processes [...], can easily be constructed on a linear network, [...]".

__

*You need sth. like an academic vpn to access and download the pdfs without any expense.

Answer 1

It would be easier to ask the authors directly.

Here is some code (written by Greg McSwiggan) to generate a Thomas cluster process on a network.

rThomaslpp <- function(L, kappa, mu, sigma) {
 X <- rpoislpp(kappa, L)
 Y <- density(X, sigma)
 Y <- eval.linim(Y * mu)
 Z <- rpoislpp(Y, L)
 return(Z)
}

The network is L. The parent intensity is kappa (points per unit length), the mean number of offspring per parent is mu (dimensionless), and the cluster size is sigma (length units).

Example:

require(spatstat)
X <- rThomaslpp(simplenet, 4, 6, 0.07)
plot(X)

This code will be added to the next version of the package spatstat.linnet.