[R] Fwd: Simulation of spatial Log-Gaussian Cox process in Spatstat

[Rolf Turner]

On 08/12/12 07:35, George Limitsios wrote:
> Hello,
>
> I have fitted a Log-Gaussian Cox Process on my data but when I try to
> use "simulate.kppm" of the spatstat package I get this error:
>
> Error in rLGCP(model = model, mu = mu, param = param, ..., win = win) :
>    The spatial domain of the pixel image ‘mu’ does not cover the
> simulation window ‘win’
>
>
> I've used covariates as im, changed the npixel value of the spatstat
> options to match the dimensions of the covariates images, but the
> problem keeps showing up.
>
> Maybe it has something to do with the irregular owin used as window of
> the ppp object?

An "irregular" window should not cause any problem.

Can you supply a reproducible example --- including a pattern
(object of class "ppp") and the image (object of class "im") that
you use as a covariate?

Then I can look into the problem.

I tried using simulate.kppm() on a toy example of a fit to an
inhomogeneous log Gaussian Cox process (with a polygonal window)
just now, and it seemed to work just fine.

     cheers,

         Rolf Turner