RELEASE NOTES

spatstat 1.38-0

15 August 2014

We thank Malissa Baddeley, Colin Beale, Oscar Garcia, Daniel Esser, David Ford, Eric Gilleland, Andrew Hardegen, Philipp Hunziker, Abdollah Jalilian, Tom Lawrence, Lore De Middeleer, Robin Milne, Mike Porter, Suman Rakshit, Pablo Ramon, Jason Rudokas, Christopher Ryan, Dominic Schuhmacher, Medha Uppala and Rasmus Waagepetersen for contributions.

OVERVIEW

spatstat now Requires the package ‘goftest’
and Suggests the package ‘Matrix’.
New dataset: ‘sporophores’
Palm likelihood method for fitting cluster processes and Cox processes.
Quasi-likelihood and weighted composite likelihood methods
for estimating trend in cluster processes and Cox processes.
Further extensions to model formulas in ppm and kppm.
Faster variance calculations for ppm objects.
One-sided tests and one-sided envelopes of summary functions.
Cramer-Von Mises and Anderson-Darling tests of spatial distribution.
Cressie-Read test statistic in quadrat counting tests.
Spatial cumulative distribution functions.
Faster algorithm for point pattern matching.
Improvements to plots.
Increased support for envelopes.
New generic functions ‘Window’, ‘Frame’ and ‘domain’.
Directional K-function and directional distribution.
Raster calculations accelerated.
Summary functions accelerated.
Many improvements and bug fixes.
Version nickname: ‘Wicked Plot’

NEW DATASETS

sporophores
Spatial pattern of three species of mycorrhizal fungi around a tree.
[Contributed by E. David Ford.]

NEW FUNCTIONS

improve.kppm
Re-estimate the trend in a kppm (cluster or Cox) model
using quasi-likelihood or weighted first-order composite likelihood.
[Contributed by Abdollah Jalilian and Rasmus Waagepetersen.]
Window, Window<-
Generic functions to extract and change the window of a spatial object
in two dimensions. Methods for ppp, psp, im, and many others.
Frame, Frame<-
Generic functions to extract and change the containing rectangle (‘frame’)
of a spatial object in two dimensions.
domain
Generic function to extract the spatial domain of a spatial object
in any number of dimensions.
Ksector
Directional version of the K-function.
pairorient
Point pair orientation distribution.
spatialcdf
Compute the spatial cumulative distribution of a spatial covariate,
optionally using spatially-varying weights.
cdf.test
[Supersedes ‘kstest’.]
Test of goodness-of-fit of a Poisson point process model.
The observed and predicted distributions of the values of a
spatial covariate are compared using either the
Kolmogorov-Smirnov, Cramer-Von Mises or Anderson-Darling test.
berman.test
Replaces ‘bermantest’.
harmonise.fv
Make several functions compatible.
simulate.lppm
Simulate a fitted point process model on a linear network.
subset.ppp, subset.lpp, subset.pp3, subset.ppx
Methods for ‘subset’, for point patterns.
closepairs.pp3, crosspairs.pp3
Low-level functions to find all close pairs of points in three dimensions
volume.linnet
Method for the generic ‘volume’. Returns the length of the linear network.
padimage
Pad the border of a pixel image.
as.layered
Convert spatial data to a layered object.
panel.contour, panel.image, panel.histogram
Panel functions for ‘pairs’ plots.
range.fv, min.fv, max.fv
Range, minimum and maximum of function values.

SIGNIFICANT USER-VISIBLE CHANGES

ppm.formula
The left hand side of the formula can now be the name of
an object in the list ‘data’, or an expression involving such objects.
ppm
The right hand side of the formula can now include
the symbol ‘.’ representing all covariates in the list ‘data’.
ppm
New argument ‘subset’ makes it possible to fit the model
in a subset of the spatial domain defined by an expression.
kppm
New option: method=”palm”, will fit the model by maximising Palm likelihood.
pppdist
Substantially accelerated.
New argument ‘auction’ controls choice of algorithm.
rhohat
New arguments ‘weights’ and ‘horvitz’ for weighted calculations.
persp.im
Surface heights and colours can now be controlled by different images.
Option to draw a grey apron around the sides of the perspective plot.
Return value has a new attribute ‘expand’.
plot.listof
New arguments ‘halign’ and ‘valign’ give improved control
over the alignment of panels.
plot.listof
If all components of the list are objects of class ‘fv’
representing functions, then if equal.scales=TRUE, these
functions will all be plotted with the same axes
(i.e. the same xlim and the same ylim).
envelope
The argument ‘transform’ is now processed by ‘with.fv’
giving more options, such as ‘transform=expression(. - r)’
envelope, dclf.test, mad.test
One-sided tests and one-sided envelopes can now be produced,
by specifying the argument ‘alternative’.
dclf.test, mad.test
A pointwise test at fixed distance ‘r’ can now be performed
by setting rinterval = c(r,r).
envelope
New arguments ‘fix.n’ and ‘fix.marks’ for envelope.ppp and envelope.ppm
make it easy to generate simulated patterns conditional on the
total number of points, or on the number of points of each type.
quadrat.test
Can now calculate the Cressie-Read test statistic
instead of the Pearson X2 statistic.
Kres, Gres, Kcom, Gcom, psst, psstA, psstG
New argument ‘model’ makes it easier to generate simulation envelopes
of the residual summary functions.
layered, plot.layered
The layer plot arguments can include the argument ‘.plot’
specifying a function to perform the plotting instead of the
generic ‘plot’.
deriv.fv
New arguments make it possible to differentiate a periodic function.
ppm
Argument ‘data’ or ‘covariates’ can now include entries
which are not spatial covariates, provided they do not appear
in the model formula.
closepairs, crosspairs
These functions are now generic, with methods for ‘ppp’ and ‘pp3’
rLGCP
Updated to conform to new usage of RandomFields package.
Argument syntax has changed.
Now allows control over pixel resolution.
bw.diggle
New arguments ‘correction’ and ‘hmax’ for controlling the calculation.
predict.lppm
New argument ‘new.coef’ for computing predictions with a
different vector of model coefficients.
predict.ppm
If ‘locations’ is a pixel image, its pixels determine the
spatial locations for prediction.
cut.ppp
Argument ‘z’ can now be a window.
split.ppp
Argument ‘f’ can now be a window.
print.ppm, summary.ppm, coef.summary.ppm
The table of parameter estimates, standard errors and confidence intervals
now also includes the value of the (signed square root) Wald test statistic.
plot.im
Now automatically detects problems in some Windows graphics displays
and tries to avoid them.
plot.im
The position of axis tick marks alongside the colour ribbon
can now be controlled using the parameter ‘at’ in the argument ‘ribargs’.
plot.ppp
Can now plot numeric marks using characters chosen by ‘pch’ or ‘chars’
with size determined by mark value.
plot.ppp
New argument ‘meansize’ for controlling mark scale.
hsvim, rgbim
New argument ‘autoscale’ causes automatic scaling of colour channel values.
plot.ppp
If type=’n’, a legend is now displayed when x is a marked point pattern.
whist
Accelerated by a factor of 5.
Fest, Jest
Accelerated by a factor of 2 to 3.
fryplot
Accelerated.
Now displays a legend if the point pattern is marked.
Now handles numerical marks nicely.
New argument ‘axes’.
frypoints
Accelerated.
New arguments ‘to’, ‘from’ and ‘dmax’.
duplicated.ppp
New option: rule = ‘unmark’
rjitter
Argument ‘radius’ now has a default.
Smooth.msr
New argument ‘drop’
LambertW
Now handles NA and infinite values.
update.ppm
Now handles formulae with a left-hand side.
raster.x, raster.y, raster.xy
These functions can now handle images, as well as masks.
Smooth.ppp
If the mark values are exactly constant, the resulting smoothed values
are now exactly constant.
eval.im, eval.fv, eval.fasp
Argument ‘envir’ can now be a list, instead of an environment.
plot.ppp
The printout (of the resulting symbol map object)
now displays the numerical value of the mark scale.
with.fv
Improved mathematical labels.
plot.fv
Improved mathematical labels on x axis.
ppm
Improved error messages.
vcov.ppm
Computations greatly accelerated for Hybrid interactions
and for Area-interaction models.
vcov.kppm
Computations greatly accelerated (when fast=TRUE)
interp.im
Argument ‘x’ can now be a point pattern.
pool.envelope
Improved handling of text information.
miplot
Improved layout.
print.summary.ppp
Improved layout.
Now respects spatstat.options(‘terse’)
print.profilepl
Improved layout.
Now respects spatstat.options(‘terse’)
anova.ppm
Now respects spatstat.options(‘terse’)
print.fv, print.envelope
Now respect spatstat.options(‘terse’) and options(‘width’)
summary.envelope
Now respects options(‘width’)
kstest, bermantest
These functions will soon be Deprecated.
They are retained only for backward compatibility.

BUG FIXES

vcov.ppm
Sometimes gave wrong answers for Poisson models fitted by method=’logi’.
Fixed.
unnormdensity
If weights were missing, the density was normalised,
contrary to the documentation.
Fixed.
logLik.ppm, anova.ppm, AIC
For models fitted by ‘ippm’, the number of degrees of freedom was incorrect.
Fixed.
im.apply
Pixels outside the window were not assigned the value NA as they should.
Fixed.
pixellate.owin
Crashed, unpredictably, if the pixel raster had unequal numbers
of rows and columns.
[Spotted by Rasmus Waagepetersen.]
Fixed.
vcov.ppm
Crashed for pairwise interaction models fitted by method=”logi”.
Fixed.
predict.ppm
Crashed for models fitted by method=”logi”
if the model included external covariates.
Fixed.
predict.ppm
Crashed if the argument ‘covariates’ or ‘data’ in the original call to ‘ppm’
included entries that were not spatial covariates. [These entries were
ignored by ppm but caused predict.ppm to crash.]
Fixed.
simulate.kppm, rNeymanScott, rThomas, rMatClust
Crashed randomly when simulating an inhomogeneous model.
[Spotted by Philipp Hunziker.]
Fixed.
bw.diggle
In some extreme cases, generated an error message
about `NaN values in Foreign function call.’
[Spotted by Colin Beale.]
Fixed.
textureplot
Crashed if ‘spacing’ was too large.
Fixed.
superimpose.psp
Crashed if the result was empty.
Fixed.
istat
Crashed with an error message about ‘vars’.
Fixed.
dirichlet, delaunay, delaunay.distance
Crashed in rare cases due to a problem in package ‘deldir’.
[Spotted by Pierre Legendre.]
Fixed.
rgbim, hsvim
Crashed if any argument was constant.
Fixed.
scaletointerval
Crashed if x was constant.
Fixed.
linnet, [.linnet
Crashed if the result contained only a single vertex.
[Spotted by Daniel Esser.]
Fixed.
plot.fv
If some of the function values were NA,
they were replaced by fictitious values
(by linearly interpolating).
Fixed.
crossdist.ppp
Ignored argument ‘squared’ if periodic=FALSE.
[Spotted by Mike Porter.]
Fixed.
marks<-.ppp
Ignored argument ‘drop’.
[Spotted by Oscar Garcia.]
Fixed.
update.ppm
Sometimes did not respect the argument ‘use.internal’.
Fixed.
plot.rhohat
Did not respect the argument ‘limitsonly’.
Fixed.
contour.im
Argument ‘axes’ defaulted to TRUE, but FALSE was intended.
Fixed.
print.hyperframe, as.data.frame.hyperframe
Column names were mangled if the hyperframe had a single row.
Fixed.
as.psp.data.frame
Generated a warning about partially-matched names in a data frame.
[Spotted by Eric Gilleland.]
Fixed.
plot.leverage.ppm
Generated a warning from ‘contour.default’
if the leverage function was constant.
Fixed.
plot.diagppm
Issued warnings about unrecognised graphics parameters.
Fixed.
update.symbolmap
Discarded information about the range of input values.
Fixed.
plot.fv
Label for y axis was garbled, if argument ‘shade’ was given.
Fixed.
plot.ppp
The legend was sometimes plotted when it should not have been
(e.g. when add=TRUE).
Fixed.
plot.listof, plot.im
In an array of plots, containing both images and other spatial objects,
the titles of the panels were not correctly aligned.
Fixed.
plot.tess, plot.quadratcount
Ignored arguments like ‘cex.main’.
Fixed.
iplot
Navigation buttons (Left, Right, Up, Down, Zoom In, Zoom Out)
did not immediately refresh the plot.
Fixed.
iplot.layered
Reported an error ‘invalid argument type’ if all layers were deselected.
Fixed.

Release notes are available in raw text format here.

Spatstat

analysing spatial point patterns