29.01.2015 Views

Analysing spatial point patterns in R - CSIRO

Analysing spatial point patterns in R - CSIRO

Analysing spatial point patterns in R - CSIRO

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

26.3 Pairwise <strong>in</strong>teraction models 157<br />

The <strong>in</strong>homogeneous Poisson process <strong>in</strong> W with <strong>in</strong>tensity function λ(u) has probability density<br />

n∏<br />

f(x) = α λ(x i ). (33)<br />

i=1<br />

where the constant α is<br />

[∫ ]<br />

α = exp (1 − λ(u))du .<br />

W<br />

The densities (32) and (33) are products of terms associated with <strong>in</strong>dividual <strong>po<strong>in</strong>t</strong>s x i . This<br />

reflects the conditional <strong>in</strong>dependence property (PP4) of the Poisson process.<br />

26.3 Pairwise <strong>in</strong>teraction models<br />

In order to construct <strong>spatial</strong> <strong>po<strong>in</strong>t</strong> processes which exhibit <strong>in</strong>ter<strong>po<strong>in</strong>t</strong> <strong>in</strong>teraction (stochastic<br />

dependence between <strong>po<strong>in</strong>t</strong>s), we need to <strong>in</strong>troduce terms <strong>in</strong> the density that depend on more<br />

than one <strong>po<strong>in</strong>t</strong>. The simplest are pairwise <strong>in</strong>teraction models, which have probability densities<br />

of the form<br />

⎡ ⎤ ⎡ ⎤<br />

n(x)<br />

∏<br />

f(x) = α ⎣ b(x i ) ⎦ c(x i ,x j ) ⎦ (34)<br />

i=1<br />

⎣ ∏ i r<br />

0 if ||u − v|| ≤ r<br />

(35)<br />

where ||u − v|| denotes the distance between u and v, and r > 0 is a fixed distance, then the<br />

density becomes<br />

{<br />

αβ<br />

n(x)<br />

if ||x<br />

f(x) =<br />

i − x j || > r for all i ≠ j<br />

0 otherwise<br />

This is the density of the Poisson process of <strong>in</strong>tensity β <strong>in</strong> W conditioned on the event that no<br />

two <strong>po<strong>in</strong>t</strong>s of the pattern lie closer than r units apart. It is known as the (classical) hard core<br />

process.<br />

Copyright<strong>CSIRO</strong> 2010

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!