Spatstat

analysing spatial point patterns

• Home
• News
• Download
• Docs
• Book
• Courses
• Help
• FAQ
• About

RELEASE NOTES

spatstat 1.52-0

10 August 2017

We thank Nicholas Read, Abdollah Jalilian, Suman Rakshit, Dominic Schuhmacher and Rasmus Waagepetersen for contributions.

OVERVIEW

• Important bug fixes.

• Now handles disconnected linear networks.

• Effect function is now available for all types of fitted model.

• A model can be fitted or re-fitted to a sub-region of data.

• More support for measures.

• ‘Pool’ operations improved.

• Geometric-mean smoothing.

• Changed algorithm defaults in ippm.

• Version nickname: “Rudimentary Lathe”

NEW FUNCTIONS

• as.data.frame.envelope
  Extract function data from an envelope object,
  including the functions for the simulated data (‘simfuns’)
  if they were saved.

• is.connected, is.connected.default, is.connected.linnet
  Determines whether a spatial object consists of
  one topologically connected piece, or several pieces.

• is.connected.ppp
  Determines whether a point pattern is connected after
  all pairs of points closer than distance R are joined.

• hist.funxy
  Histogram of values of a spatial function.

• model.matrix.ippm
  Method for ‘model.matrix’ which allows computation of
  regular and irregular score components.

• harmonise.msr
  Convert several measures (objects of class ‘msr’)
  to a common quadrature scheme.

SIGNIFICANT USER-VISIBLE CHANGES

• Smooth.ppp
  New argument ‘geometric’ supports geometric-mean smoothing.

• Kinhom
  New argument ‘ratio’.

• linearKinhom, linearpcfinhom
  Changed default behaviour when ‘lambda’ is a fitted model.
  New arguments ‘update’ and ‘leaveoneout’.

• linearK, linearKinhom, linearpcf, linearpcfinhom, compilepcf
  Ratio calculations are now supported.
  New argument ‘ratio’.

• effectfun
  Now works for ‘ppm’, ‘kppm’, ‘lppm’, ‘dppm’, ‘rppm’ and ‘profilepl’ objects.

• ppm, kppm
  The argument ‘subset’ can now be a window (class ‘owin’)
  specifying the subset of data to which the model should be fitted.

• fitted.lppm
  New argument ‘leaveoneout’ allows leave-one-out computation of fitted value.

• pool.rat
  New arguments ‘relabel’ and ‘variance’.

• density.lpp
  The return value is a pixel image of class ‘linim’ in all cases.

• plot.linim, plot.linfun
  A scale bar is now plotted when style=”width”.
  New argument ‘legend’.

• ippm
  Default values for the parameters of the optimisation algorithm (nlm.args)
  have changed.

• ippm
  The internal format of the result has been extended slightly.

• bind.fv
  New argument ‘clip’.

• as.im.distfun
  New argument ‘approx’ specifies the choice of algorithm.

• ”[.psp”
  New argument ‘fragments’ specifies whether to keep fragments of
  line segments that are cut by the new window, or only to retain
  segments that lie entirely inside the window.

• predict.rhohat
  New argument ‘what’ determines which value should be calculated:
  the function estimate, the upper/lower confidence limits, or the
  standard error.

• pool.fv
  New arguments ‘relabel’ and ‘variance’

• pool.rat
  New argument ‘weights’.

• plot.msr
  New argument ‘massthresh’.

• Ops.msr
  Calculations like A+B can now be performed even when the measures A and B
  are not defined on the same quadrature scheme.

• density.ppp
  New argument ‘verbose’.

• bw.pcf
  New argument ‘verbose’.

• hist.im
  New argument ‘xname’.

• [.leverage.ppm
  New argument ‘update’.

• [.layered
  Additional arguments ‘…’ are now passed to other methods.

• logLik.ppm
  The warning about pseudolikelihood (‘log likelihood not available’)
  is given only once, and is not repeated in subsequent calls,
  within a spatstat session.

• kppm
  Refuses to fit a log-Gaussian Cox model with anisotropic covariance.

• plot.linim, plot.linfun
  The return value has a different format.
  Arguments have been renamed and reorganised.

• density.lpp
  New argument ‘old’.

• ippm
  Accelerated.

• Smooth.ppp
  Now exits gracefully if any mark values are NA, NaN or Inf.

• timeTaken
  Now exits gracefully if there is no timing information.

• nbfires
  The unit of length for the coordinates is now specified in this dataset.

BUG FIXES

1. bw.pcf
   Results were totally incorrect due to a typo.
   [Spotted by Abdollah Jalilian and Rasmus Waagepetersen.]
   Fixed.

2. predict.rho2hat
   Results were incorrect for a rho2hat object computed from a point pattern.
   Fixed.

3. density.ppp
   If the smoothing bandwidth was very small (e.g.\ smaller than pixel width),
   results were inaccurate if the default resolution was used,
   and completely wrong if another resolution was specified.
   [Spotted by Dominic Schuhmacher.]
   Fixed.

4. linearK, linearKinhom, linearpcf, linearpcfinhom, linearKcross,
   linearKdot, linearpcfcross, linearpcfdot, linearKcross.inhom,
   linearKdot.inhom, linearpcfcross.inhom, linearpcfdot.inhom
   Crashed if the network was disconnected.
   Fixed.

5. crossdist.lpp
   Crashed if the network was disconnected.
   Fixed.

6. countends
   Crashed if the network was disconnected.
   Fixed.

7. model.images.ppm
   Crashed for models fitted using ‘covfunargs’.
   Fixed.

8. model.matrix.ppm
   Crashed for models fitted using ‘covfunargs’,
   if argument ‘Q’ was given.
   Fixed.

9. polynom
   Expansion of some polynomials caused an error message
   about ‘invalid model formula’.
   Fixed.

10. plot.ppp
    The argument ‘type=”n”’ did not suppress plotting of the legend,
    for marked point patterns.
    Fixed.

11. plot.psp
    Ignored ‘show.all’ when ‘add=TRUE’.
    Fixed.

12. intensity.ppm
    Result had incorrect ‘names’ attribute in some cases.
    Fixed.

13. marks<-.ppx
    The assignment marks(X) <- a, where ‘a’ is a single atomic value,
    caused an error if ‘X’ contained zero points.
    Fixed

14. model.depends
    Crashed when applied to regression models fitted by ‘gam’,
    or point process models fitted by ‘ppm’ with ‘use.gam=TRUE’.
    Fixed.

15. pool.fv
    Crashed sometimes, if the arguments did not have the same set of
    column names.
    Fixed.

16. pool.rat
    Crashed with an error message from ‘fmt’
    if there were more than 20 objects to be pooled.
    Fixed.

17. linearK
    The ‘theo’ column was missing if npoints(X) < 2 and correction=”Ang”.
    Fixed.

18. model.matrix.ppm
    Result was malformed if the model was fitted with ‘use.gam=TRUE’.
    Fixed.

19. effectfun
    Crashed if ‘covname’ was omitted,
    if the model was fitted with ‘use.gam=TRUE’.
    Fixed.

20. nncross.lpp
    Result had incorrect format if Y was empty, in some cases.
    Fixed.

21. linearKinhom
    Plot label for y axis was incorrect.
    [Spotted by Suman Rakshit.]
    Fixed.

22. plot.solist
    If the entries were ‘linim’ objects, they were plotted using image()
    so arguments like ‘style=”w”’ were ignored.
    Fixed.

23. as.ppp.data.frame
    Crashed if X was an object of class ‘tbl_df’ from the dplyr package.
    Fixed.

24. plot.lpp
    Crashed if there were multiple columns of marks.
    Fixed.

Release notes are available in raw text format here.