Spatstat

analysing spatial point patterns

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

RELEASE NOTES

spatstat 1.60-0

23 June 2019

We thank Ottmar Cronie, Tilman Davies, Andrew Hardegen, Tom Lawrence, Robin Milne, Mehdi Moradi, Gopalan Nair, Tim Pollington and Suman Rakshit for contributions.

OVERVIEW

• Random thinning and random labelling of spatial patterns
  extended to different types of pattern.

• Confidence intervals for multitype K functions.

• Envelopes for balanced two-stage test

• Accelerated some code.

• Minor bug fixes and improvements.

• Package built under R 3.6.0.

• Version nickname: ‘Swinging Sixties’

NEW FUNCTIONS

• bits.envelope
  Global simulation envelope corresponding to bits.test,
  the balanced independent two-stage Monte Carlo test.

• extrapolate.psp
  Extrapolate line segments to obtain infinite lines.

• uniquemap
  Map duplicate points to unique representatives.
  Generic with methods for ppp, lpp, ppx

• uniquemap.data.frame
  Map duplicate rows to unique representatives

• localKcross, localLcross, localKdot, localLdot,
  localKcross.inhom, localLcross.inhom
  Multitype local K functions.

SIGNIFICANT USER-VISIBLE CHANGES

• lohboot
  Now works for multitype K functions Kcross, Kdot, Lcross, Ldot,
  Kcross.inhom, Lcross.inhom.

• Kinhom, pcfinhom
  Leave-one-out calculation is implemented when ‘lambda’ is a fitted model
  of class ‘dppm’.

• Kcross.inhom, Kdot.inhom, Lcross.inhom, Ldot.inhom
  Leave-one-out calculation is implemented when ‘lambdaX’ is a fitted model
  of class ‘dppm’.

• rthin, rlabel
  These functions now work on point patterns of all classes
  (ppp, lpp, pp3, ppx) and line segment patterns (psp).

• bw.abram
  New argument ‘smoother’ determines how the pilot estimate is computed.
  Formal arguments rearranged.

• plot.im
  New argument ‘riblab’.

• rlabel
  New arguments ‘nsim’ and ‘drop’.

• localK, localKinhom
  New argument ‘rmax’.

• rLGCP
  Accelerated.

• anyDuplicated.ppp
  Accelerated.

• duplicated.ppp
  Accelerated, in most cases.

• simulate.kppm
  Accelerated, for LGCP models.

• predict.ppm
  Accelerated, for models fitted with method=”VBlogi”

• print.rmhmodel
  Output improved.

BUG FIXES

1. plot.linim, plot.linfun
   Not all the entries of ‘leg.args’ were passed to text.default.
   Fixed.

2. densityVoronoi.ppp
   Did not correctly handle patterns containing duplicated points.
   Fixed.

3. markcorr
   The argument ‘correction=”none”’ did not generate any results.
   [Spotted by Tim Pollington.]
   Fixed.

4. names<-.fv
   Did not adjust the plotting formula, so that a subsequent call to plot.fv
   would complain about missing variables.
   Fixed.

5. im.apply
   Crashed if ‘FUN’ returned factor values.
   Fixed.

6. stienenSet
   Crashed if the data contained duplicated points.
   Fixed.

7. predict.ppm, effectfun
   Crashed in some cases, with the message ‘ncol(x)=nrow(v) is not TRUE’.
   Fixed.

8. parres
   Crashed in some cases, with the message ‘logical subscript too long’.
   Fixed.

9. dclf.test, mad.test, dg.test, bits.test
   Crashed in some cases if the summary function values were infinite or NA.
   Fixed.

Release notes are available in raw text format here.