[R-sig-Geo] qqplot.ppm

[Tom_R]

Hi List,
I am attempting to obtain a qqplot for a ppm  (point Poisson model) object.
The object is a Geyer model ('Hard0'). The 'Saturation' and 'Interaction
range' parameters for the Geyer model are obtained   using profile
log-likelihood ('profilepl').

I am using a spatial covariate (Z) which is a pixel image (dens_all). This
covariate is causing problems for qqplot.ppm. I am getting the following
error from qqplot.ppm; "ERROR : Error in rmhResolveExpansion(win, control,
covims, "covariate") : 
  Cannot expand the simulation window, because the covariate images do not
cover the expanded window". 

My code is below.

So, dear list, any suggestions on how to correct the qqplot.ppm error will
be most appreciated!!!
Many thanks in advance,
Tom Richardson


##########################################################
##########################################################

for (i in 1:length(unique(colony)))
  {

  for (j in 1:  (length(unique(photo[colony==i])))  )	## '-1' prevents any
crossover between 2 colonies
 	{

wind       <- owin(c(xmin , xmax ), c(ymin ,ymax  ) )	## a window 
points     <- ppp(x=x[colony==i & photo==j],y=y[colony==i & photo==j], 
window= wind)	## Point pattern for a single photo
all_points <- ppp(x=x[colony==i], y=y[colony==i],  window= wind)  ## A ppp
for all photos overlaid

## intensity field; calculated over  all photos superimposed
dens_all    <- as.im(density(all_points, kernel = "epanechnikov",correction
= c("translate")))

## use profile log-likelihood to select irregular parameters for the Geyer
model:
s                  <- expand.grid( r=seq(1,3, by=0.25), sat=seq(1,3,by=1))	
pg                 <- profilepl(s, Geyer, points)		
GAMMA 		   <- exp(pg$fit[6]$fitin$coefs[2])		#Gamma > 1: clustering, <1:
regularity
INTERACTION_RANGE  <- pg$fit$interaction$par$r  
SATURATION	   <- pg$fit$interaction$par$sat

## Create a non-stationary  model of the point process with the pixel image,
'dens_all' as a covariate.
Hard0      <- ppm(points, ~ polynom(x,y,2) + Z, covariates=list(Z=dens_all),
Geyer(INTERACTION_RANGE ,SATURATION))

## Diagnostic plots for the model; 
diagnose.ppm(Hard0)
qqplot.ppm(Hard0, fast=TRUE, plot.it=FALSE, nsim=50)	        ## Now check
the interpoint interaction 

## loops finish
        }
 }

##########################################################
##########################################################
-- 
View this message in context: http://r-sig-geo.2731867.n2.nabble.com/qqplot-ppm-tp5555735p5555735.html
Sent from the R-sig-geo mailing list archive at Nabble.com.