RELEASE NOTES 

		spatstat 1.5-2

	 	30 July 2004


	We thank Stephen Eglen and Evgueni Parilov
	for helpful contributions.

OVERVIEW

	We have improved the code for
	prediction, simulation and plotting of fitted models.
	There are minor changes to usage.

	Fitted models with the PairPiece interaction
	can now be simulated using rmh.ppm.

	There are some new facilities for basic handling
	of point patterns.

	Numerous bugs have been fixed.

COMPATIBILITY

	1. In rmh.default with cif="lookup", the usage of the
	arguments par$r and par$h has changed. Additionally, par$r
	is now required to be sorted in ascending order, and to have r[1] > 0.
	See help(rmh.default).

	2. In PairPiece interaction models, the pair interaction function 
	is now assumed to be right-continuous, not left-continuous.

	3. The default behaviour of predict.ppm, when neither the grid
	dimensions nor the prediction locations are specified, 
	has been altered so that it predicts on a 50 x 50 grid of locations.

NEW STUFF

	1. New dataset 'betacells' (thanks to Stephen Eglen). 
	This is a corrected version of the dataset 'ganglia'
	which has been found to contain errors.
	See the help files for 'betacells' and 'ganglia'.

	2. New function "[<-.ppp" allows you to 
	replace subsets of a point pattern, as in
		X[subset] <- Y

	3. New function 'split.ppp', the analogue of 'split' 
	for point patterns. Useful for separating the points 
	of different types in a multitype pattern.

	4. Associated plot method "plot.splitppp". 
	   Try
		plot(split(amacrine))
	

IMPROVEMENTS/ENHANCEMENTS

	1. rmh.ppm will now simulate fitted models
	with the PairPiece interaction (arbitrary pairwise
	interaction processes with isotropic pair potential)

	2. In rmh.default with cif="lookup", the interaction
	may be supplied as a 'stepfun' object, instead of 
	as a pair of vectors.

	3. plot.ppm now returns an object of class "plotppm"
	which can be saved and re-plotted with other graphics settings
	without having to re-compute the coordinates of the surfaces.
	See help(plot.ppm)

	4. In plot.ppm and predict.ppm, the argument 'locations' 
	may now be a mask   (an owin object of type "mask").

	5. as.mask now accepts x, y coordinates of the pixel grid,
	allowing complete control over digitisation of windows
	and over the locations for predicting a fitted model.

	6. More features have been added to 'superimpose'.
	For example
		superimpose(a=rpoispp(10), b=rpoispp(20))
	will produce a multitype pattern with types "a" and "b".

	7. The appearance of plots made by plot.ppm has improved slightly.

	8. demo(spatstat) has been expanded slightly.

	9. The plotting functions for "ppm" and "im" objects
	(plot.plotppm, plot.im, contour.im, persp.im, image.im)
	now have better handling of the generic plot arguments
	such as 'main', 'xlab', 'ylab'. 

	10. In rmh.ppm there are now default values for 'start' 
	and 'control' so that you can type
		fit <- ppm(.....)
		rmh(fit)

BUGS

	1. Two bugs in predict.ppm caused it to crash sometimes
	when the fitted model was a multitype process
	with covariates. Fixed.

	2. A bug in predict.ppm caused it to give incorrect answers
	sometimes when the predicted potential is -Infinity. 
	Fixed.

	3. A bug caused rmh.default to crash sometimes
	with cif="lookup". Noticed by Evgueni Parilov.
	Fixed.

	4. A bug caused rmh.ppm to crash sometimes
	when trying to simulate a fitted model
	with the MultiStrauss or MultiStrHard interactions. 
	Fixed.

	5. A minor bug affecting the appearance of 
	contour plots in contour.im has been fixed.

	6. Minor bugs in examples(ppm.object) and examples(ppm) 
	have been fixed.