Likelihood and Non‐parametric Bayesian MCMC Inference for Spatial Point Processes Based on Perfect Simulation and Path Sampling (Q4455960)

A pointwise interaction point process on a bounded planar region \(S\) is considered with density w.r.t. the unit rate Poisson process defined by \[ f(x)=c_{\beta,\varphi}^{-1} \beta^{\text{\,card}(x)} \prod_{\{\xi,\eta\}\subseteq x:\xi\not=\eta} \phi(\| \xi-\eta\|), \] where \(0\leq\varphi\leq 1\) is an interaction function, \(x\) denotes any finite subset of \(S\), \(\beta>0\) is a parameter, and \(c_{\beta,\varphi}\) is a normalizing constant. Approximate MLE estimation is described for \(\varphi(r)=\gamma^{1\{r\leq R\}}\) (a Strauss process). A nonparametric Bayesian analysis technique is developed for \(\phi\) approximated by a step function. The perfect simulation with dominated coupling from the past (CFTP) algorithm is used for the sampling.