- Following R fixed bug [PR#18151] (https://bugs.r-project.org/show_bug.cgi?id=18151=),
`mhat()`

code is modified by Martin Maechler.

- An error occurred when a function was called with
`dbmss::`

prefix and`CheckArguments = TRUE`

and dbmss was loaded.

`fmla`

argument added in`autoplot.envelope()`

.`autoplot.fv()`

added.`autoplot.wmppp()`

added.- CodeFactor optimization applied.
- Suggests rmarkdown (https://github.com/yihui/knitr/issues/1864)

- Continuous integration by GitHub Actions.

- Updates in the
*spatstat*package:*dbmss*has been updated to address the creation of*spatstat.core*et al.

- argument
`main`

in`autoplot()`

was ignored.

- tibbles are accepted by
`wmppp()`

. - Precision of point coordinates can be set in
`rRandomPositionK()`

.

`autoplot()`

failed with`Lenvelope`

objects.

`autoplot()`

for`envelope`

objects.- Vignettes by pkgdown
- Better consistency of bandwidth choice in
`Kd`

and`m`

estimation. An article is added in the online documentation.

- Fewer imported functions in NAMESPACE.

`Kdhat()`

density was normalized incorrectly with Approximate=TRUE and small r values.

- More robust check of arguments.
- Rcpp >= 0.12.14 required.
`init.c`

no longer necessary. - Introduction vignette.
- On Travis and codecov.io now.
- NEWS in Markdown.
`wmppp()`

accepts characters in`PointType`

. A`PointName`

column is searched to set the row names of the marks.

`wmppp()`

failed if some point laid outside the window.

- On GitHub now.
- Individual values of m and M available from
`Mhat`

and`mhat`

with argument`Individual`

. - Point names can be specified as row names of the dataframe passed to
`wmppp()`

. They are preserved in the`fv`

object returned by individual`Mhat`

and`mhat`

to identify points more easily.

- Far less memory is used to compute M and approximated m functions. The ratio of necessary memory equals that of the number of reference points to the total number of points.
`[.wmppp`

function added to*spatstat*generics

- Declaration of the required versions of R and
*spatstat*in DESCRIPTION to avoid error in CRAN test on oldrel Windows platform.

- Updates in the
*spatstat*package:*dbmss*has been updated to address the creation of*spatstat.util*.

- C routines registration to comply with R 3.4 policy.
- Explicit export of all non-internal functions instead of
`exportPattern("^[[:alpha:]]+")`

. - Import of package
*cubature*reduced to function`adaptIntegrate()`

.

- Enforcement of the use of C++11 to avoid warnings due to
*RcppParallel*.

- Distance matrices can be used instead of point patterns in
`Mhat`

,`mhat`

and`Kdhat`

and their envelopes. - The estimation of density used by
`Kd`

now includes reflection: the estimation of`Kd`

is more accurate than before, but values may vary from previous versions close to the minimum distance. - LazyData is used to save memory.

- Weighted argument was ignored in
`Kdhat`

(bug introduced in v.2.2-4). Fixed.

`Mhat`

and`mhat`

C++ code is now parallelized thanks to*RcppParallel*.- Small performance improvement of
`Kdhat`

(around 5% faster).

- Updated CITATION: the paper about this package has been published: Eric Marcon, Stephane Traissac, Florence Puech, Gabriel Lang (2015). Tools to Characterize Point Patterns: dbmss for R. Journal of Statistical Software, 67(3), 1-15.

- Intervals of estimation of the Kd function are narrower when the distance range is small with respect to the window to improve precision.

- Distances pairs more than twice the maximum value of
`r`

where ignored when using the approximate estimation of`Kd`

, so`Kd`

was overestimated when`r`

was smaller than usual (much less than the default values). Corrected.

- Adjust argument ignored in
`Kdhat`

. Fixed.

`m`

function added.- Default value
`Controls=NULL`

for`Dhat`

: controls are by default all points except for cases. `Kd`

and`m`

are both computed by default up to one third of the diameter of the window, other options are added.`Original`

and`Adjust`

arguments allow to change the default bandwith used by`Kd`

and`m`

.

- Typo in internal function name
`FillEnveloppe`

corrected:`FillEnvelope`

is the new name.

- Default neighbor type for
`Khat`

and`Kenvelope`

is the same as reference type instead of "". `Kd`

estimation now accepts "" as reference type to use all points.- Approximate argument to calculate
`Kd`

on big data sets with little RAM.

`FillEnvelope`

returned wrong quantiles for local confidence intervals. Fixed.

- Global confidence intervals are now defined even when
`NA`

values are found in the simulations. `Kd`

is computed by default up to the median (instead of mean) distance between points.

`Kdhat`

with Reference and Neighbor points of different types was calculated with only half of point pairs and sometimes crashed R. Fixed.

`Kdhat`

and`M`

are computed through C++ loops instead of spatstat`pairdist()`

. Much faster, and uses far less memory.

`Kdhat`

and`M`

propose a default value for argument`r`

.- Vignette added.

`Kdhat`

probability density estimation is cut at the lowest distance between points instead of 0.

`Kdhat`

with Reference and Neighbor points of the same type and`Weighted=TRUE`

returned an error. Fixed.

- Minor corrections, mainly comments and formating.

- Most code rewritten.

- Names of functions
`X.r`

became Xhat. - Point patterns are now of class
`wmppp`

. - Results are of class
`fv`

, envelopes of class`envelope`

.

- NEWS file added.

- Documentation format improved.

- The default behavior of
`Kd`

is that of Duranton and Overman (2005). An optional parameter is added to used improved bandwith selection.

- The density estimation of
`Kd`

was run twice. Fixed.

- First version on CRAN. Versions 1.2.2 and 1.2.3 contain faster examples to follow CRAN requirements.

`Kinhom`

function added.

- First version.