RELEASE NOTES

spatstat 1.57-0

30 October 2018

We thank Corey Anderson, Ryan Arellano, Hadrien Commenges, Ottmar Cronie, Tilman Davies, Maximilian Hesselbarth, Kassel Hingee, Tomas Lazauskas, Marie-Colette van Lieshout, Rasmus Waagepetersen and ‘daitakahashi’ for contributions.

OVERVIEW

Cronie-van Lieshout bandwidth selection.
Smoothing functions handle non-Gaussian kernels.
Infinite smoothing bandwidths permitted.
Positive confidence limits for rhohat.
Improved bivariate interpolation.
subset() method for line segment patterns.
Important bug fixes in rthin and density.ppp
Minor bug fixes and improvements.
Version nickname: ‘Zombie apocalypse’

NEW FUNCTIONS

bw.CvL
Cronie-Van Lieshout bandwidth selection for density estimation.
[Contributed by Ottmar Cronie and Marie-Colette van Lieshout.]
subset.psp
Method for ‘subset’ for line segment patterns.

SIGNIFICANT USER-VISIBLE CHANGES

densityfun.ppp, Smoothfun.ppp, Smooth.ppp
These commands now handle non-Gaussian kernels.
density.ppp, relrisk.ppp, Smooth.ppp, densityfun.ppp, Smoothfun.ppp
Argument ‘sigma’ can be infinite.
interp.im
New argument ‘bilinear’ specifies the choice of interpolation rule.
rhohat
Methods for rhohat have a new argument ‘positiveCI’
specifying whether confidence limits should be positive.
plot.colourmap
New argument ‘increasing’ specifies whether the colours are
displayed in order left-to-right/bottom-to-top.
Changed default behaviour for discrete colour maps when vertical=FALSE.
split.ppx
Argument ‘f’ can be a logical vector.
relrisk.ppp
If se=TRUE and at=”pixels”, the result belongs to class ‘solist’.
imcov, setcov, convolve.im
The name of the unit of length is preserved.
density.ppp
Slightly accelerated for non-Gaussian kernels.
bw.scott
The two bandwidth values in the result
now have names (‘sigma.x’ and ‘sigma.y’).
pairdist.default
Now checks whether the data are valid 2-dimensional coordinates.
pixellate.ppp
New argument ‘savemap’
rtemper
New argument ‘track’.
Code runs slightly faster.
eval.im, eval.linim
New argument ‘warn’.
Kres, Kcom, Gcom, psstG
If any of the calculated weights for the summary function
are infinite or NA, they are reset to zero, with a warning,
instead of a fatal error.

BUG FIXES

rthin
If P was close to 1, the result was sometimes an empty point pattern
when it should have been identical to X.
[Spotted by Maximilian Hesselbarth.]
Fixed.
density.ppp
Standard error calculations were incorrect
when sigma was a single numeric value.
The output was equal to ‘sqrt(sigma)’ times the correct answer.
Fixed.
density.ppp
Result was incorrect for non-Gaussian kernels when
at=”points” and leaveoneout=FALSE.
Fixed.
density.ppp
Did not pass additional arguments “…” to a user-supplied
kernel function, in some cases.
Fixed.
as.im.function, as.im.funxy
If the function values were factor values and the window was
not a rectangle, the result was an image with all pixel values NA.
[Spotted by Corey Anderson.]
Fixed.
plot.funxy
If the function values were factor values and the window was
not a rectangle, the plot was empty.
[Spotted by Corey Anderson.]
Fixed.
nnorient
Crashed if the border correction did not retain any data points.
[Spotted by Tomas Lazauskas.]
Fixed.
linim
Crashed in some cases with a message about unitnames.
Fixed.
density.lpp
Default value of ‘dx’ was sometimes incorrect.
Fixed.
rMatClust, rThomas, rCauchy, rVarGamma
Issued a spurious warning about bandwidth selection
when saveLambda=TRUE.
Fixed.
density.ppp
Issued a spurious warning about bandwidth selection
when ‘kernel’ was a user-supplied function.
Fixed.
clusterfield.function
Issued a spurious warning about bandwidth selection.
Fixed.
relrisk.ppp
Issued a spurious warning if the argument ‘case’ or ‘control’
was given, for a bivariate point pattern.
Fixed.
superimpose.ppp
If ‘W’ was a character string or function, the behaviour was
not exactly as described in the help file.
Fixed.
plot.psp
If the marks were factor values, the colour map was displayed upside down.
Fixed.
eval.fv
If one of the ‘fv’ objects included ratio information (class ‘rat’)
then this was erroneously retained, in some cases.
Fixed.
linearKcross
Crashed (with a message about a missing value of ‘correction’)
if there were no pairs of points to count.
Fixed.
envelope.lpp
Crashed (randomly) when fix.n=TRUE and fix.marks=TRUE.
Fixed.

Release notes are available in raw text format here.

Spatstat