RELEASE NOTES 

		spatstat 1.6-6

		10 May 2005

	We thank Marc Genton, Matt Reiter, Peter Surovy
	and Selene Wong for contributions.
	

OVERVIEW

	Point process models can now be fitted using
	approximate maximum likelihood, instead of maximum pseudolikelihood. 

	Model-fitting runs faster.
	Geometric operations on windows run faster.

	Various small bugs have been fixed.

IMPROVEMENTS/ENHANCEMENTS

	1. Huang-Ogata method
		ppm() will now fit point process models using either
		the method of maximum pseudolikelihood (method="mpl")
		or Huang-Ogata approximate maximum likelihood (method="ho").
		Maximum pseudolikelihood is much faster, but the
		Huang-Ogata method has better statistical properties.

	2. faster model-fitting
		ppm and update.ppm now run faster.
		
	3. quadrature points
		The default quadrature scheme used to fit a point process model
		is now generated in a slightly different way. This has been
		done to eliminate some minor error messages, and to accelerate
		the computation. 

	4. faster handling of windows
		Operations such as the conversion of a polygon to a mask, 
		and testing whether a point is inside a polygon, 
		now run much faster.
		Conversion of a binary mask to a binary mask of different
		dimensions also runs much faster.

	5. plotting of function arrays
		plot.fasp has been modified so that, by default,
		all the plot panels have the same x-axis limits.
		This is controlled by the argument 'samex=TRUE'.
		(Thanks to Selene Wong)

	6. example(owin) 
		now includes an example of a polygon
		with a polygonal hole. (Thanks to Matt Reiter)

NEW FUNCTIONS

	suffstat
		Computes the canonical sufficient statistics
		of a point process model.

COMPATIBILITY

	[1] The format of fitted point process models ("ppm" objects)
	has changed slightly. 

	Fitted point process models obtained using previous versions
	of spatstat cannot be modified using update.ppm(), 
	but in all other respects they are fully compatible
	with the new spatstat.

	[2] The quadrature points used to fit a point process model 
	are now generated in a slightly different way. 

	There may be slight inconsistencies in the fitted parameters
	for the same model fitted in different versions of spatstat.

BUGS

	[1] update.ppm 
		update.ppm did not handle covariates properly.
		Fixed.

	[2] ksmooth.ppp
		ksmooth.ppp crashed sometimes 
		when the argument 'dimyx' was given.
		Fixed.

	[3] quadscheme, ppm
		quadscheme() did not work correctly
		when 'ntile' was given and ntile[1] != ntile[2].
		Fixed.

	[4] markcorr
		The default plotting action 
			plot(markcorr(....))
		just displayed a horizontal line
		and did not plot the estimated function.
		Fixed.
		(Thanks to Peter Surovy)

	[5] Fest, Gest, Gcross, Gdot, Gmulti, Jest, Jdot, Jcross, Jmulti,
	  alltypes(..., "G"),  alltypes(..., "J")
		The default plot action displayed only one of the 
		function estimates. It now displays them all
		(thanks to Selene Wong)
	
	[6] Kest.fft: 
		one component of the return value 
		was incorrectly labelled. 
		Fixed.
		(Thanks to Marc Genton)

	[7] example(markcorr) 
		Running these examples took an eternity. 
		Fixed.