I have estimated the parameters of a non-homogeneous poisson space point process using Metropolis+gibbs from some observation data. I know:

$N$: Number of points inside my observation window

$\lambda$: Intensity function

and the parameters associated with this intensity function.

If I want to generate a pattern of points using the intensity function $\lambda$. This is possible with the function rpoint from the r library spatstat: https://stackoverflow.com/questions/33874879/how-to-generate-spatial-points-with-a-pattern

My question is how in general can I generate patterns of points for any distribution function assuming that I have done inference about the parameters already. I know that rpoint assumes independence between points but how can I generate a pattern of m points from a different point process, let's say the Area interaction point process assuming that I have estimated the parameters? I have read some about the reject sampling but is not straightforward to me how to do this.

By the way, the same r package has a function that brings a pattern of points for the area-interaction process but I can not specify the $m$ number of points to be generated.



I only have my phone here so my answer will be a bit rough around the edges. I believe you are looking for the function rmh.default in spatstat. To do simulation conditionally on the number of points you can set the control parameter p to 1 as described in the help file. The second example of the help file shows how to simulate a Strauss process with the number of points fixed to 42.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.