RELEASE NOTES 

		spatstat 1.17-3

		15 december 2009
	
	We thank Ege Rubak, Ya-Mei Chang, David Newbery,
	Sebastian Schutte and Karl Ove Hufthammar for contributions.

OVERVIEW
	Two important bugs fixed. 
	Point process models can now be fitted to very large 
	point pattern datasets. NA's in covariates are now handled better.
	Sundry additions and improvements.

NEW FUNCTIONS

	integral.im
		Integral of a pixel image
		(sum of pixel values * pixel area).

	pcfdot
		Multitype pair correlation (counterpart of Kdot)

	whist
		Weighted histogram

IMPROVEMENTS/ENHANCEMENTS

	ppm
		Point process models can now be fitted to
		point pattern datasets with very large numbers of points
		(and to quadrature schemes with a very large number of
		quadrature points). 
		This is done by splitting the calculation into blocks.

	predict.ppm
		It is now possible to compute predicted intensity and
		conditional intensity at a very large number of points
		(e.g. on a very fine pixel grid).
		This is done by splitting the calculation into blocks.

	spatstat.options
		New option "maxmatrix" for controlling the size of data blocks
		in ppm() and predict.ppm().

	ppm
		Improved algorithm for handling covariates
		(when they are given as pixel images).
		Warnings about NA's in the covariates 
		should happen less often.

	plot.owin
		New argument 'invert' makes it easier to invert the colours
		when displaying a binary mask.

	tess
		New argument 'keepempty' allows empty tiles 

	all summary functions (Kest, Gest, Fest, ...)
		Slightly accelerated.

	markcorr
		Inserted more checks for sanity of data.

	print.ppm
		Behaviour improved in the case where 
		the model coefficient vector has length zero.

BUGS

	[1] Kest, Kinhom, Kcross, Kdot, Kmulti, localK,
	Kcross.inhom, Kdot.inhom, Kmulti.inhom, Kest.fft

		Results may have been inaccurate
		when the observation window was a binary mask, and/or 
		when the arguments lambda,lambdaI,lambdaJ were pixel images.
		Fixed.
		
	[2] quadratcount, quadrats
		For non-rectangular windows, if xbreaks and ybreaks were 
		given, the quadrat counts were INCORRECT. 
		Fixed.

	[3] plot.ppp
		When the window was a binary mask, the points were
		sometimes invisible. 
		Fixed.