announcing secr 2.8.2

announcements (new versions, changes, bugs, installation problems...) related to program DENSITY and the R package secr

announcing secr 2.8.2

Postby murray.efford » Thu May 01, 2014 5:01 pm

A new minor release of secr is now available on CRAN. I seem to have omitted to post the last two releases, so here are their changes as well.
-- Murray

Changes in version 2.8.2 (2014-04-30)

New features or changed behaviour of old functions
o write.captures() and write.capthist() acquire new argument
tonumeric to allow output of character-valued covariates without
conversion to numeric
o region.N() saves attribute 'regionarea'
o sim.popn() accepts a vector of densities D when nsession > 1
o sim.capthist() new argument savepopn allows simulated population
to be saved as an attribute

Bugs fixed
o output from derived() misformatted when groups specified
o plot.secr failed with fixed g0 or lambda when ylim not specified

Changes in version 2.8.1 (2014-04-03)

New features or changed behaviour of old functions
o usagePlot() sums over occasions when occasion = "ALL" or
length(occasion)>1

Bugs fixed
o reduce.traps() failed when newtraps was not specified

Changes in version 2.8.0 (2014-03-31)

New features or changed behaviour of old functions
o new functions par.secr.fit(), par.region.N, and par.derived()
for parallel fit of multiple models
o Rsurface() generates a smoothed 'resource availability' surface,
with plot and print methods
o summary.mask() distinguishes Dsurfaces and Rsurfaces from other
mask types
o sim.popn() now optionally simulates a fixed population N when
model2D = "IHP" (but with Ndist = "fixed" the 2-D distribution is no
longer strictly inhomogeneous Poisson)
o rbind.capthist() for polygon and telemetry inputs automatically
forms a new search polygon that is the union of the input polygons
o new functions CV(), CVa() and CVa0(), the latter two for
summarising heterogeneity in mixture models
o rbind.traps() has new argument addusage to generate binary
usage attribute with number of occasions specific to each input
o region.N() has new argument pooled.RN for extrapolating
realised N to a landscape sampled with multiple non-overlapping
detector layouts ('sessions')
o secr.fit() allows optional parameterisations such as (D, g0,
sigmak) where sigma = sigmak / sqrt(D). See
secr-parameterisations.pdf.
o new generic function clone() with default and popn methods
(replicates points a fixed or random number of times e.g. for
generating overdispersed cues from a distribution of sources)
o trim() method for fitted models by default now also drops 'call'
component
o pdot.contour and buffer.contour() now accept multi-session input
o implementation of concurrent and independent telemetry improved,
and allows covariates; reduce.capthist() handles telemetry data
o new functions plotMCP() to display telemetry locations and
ellipse.bvn() to display confidence region for centroid.
o telemetryxy() extracts xylist attribute (telemetry coordinates)
of a combined capthist object
o addTelemetry() allows detector type "multi" and no longer
discards individual covariates from the detection capthist if
there are no all-zero histories and telemetry data does not
include covariates
o new function PG() returns proportion of telemetry fixes in a set
of polygons (i.e., 'on grid')
o sim.capthist() has new argument exactN to specify number of
telemetry fixes

Bugs fixed
o pointsInPolygon() and hence make.mask() failed with polygon
input when polygon was an sp object with attribute data (2.6 only)
o predictDsurface() and region.N() could fail or produce
improbable estimates when the user fixed beta parameters with
details$fixedbeta

Known bugs
o predict.secr() fails when newdata includes only one level for h2
under the 'hcov' formulation
o sim.capthist() does not use a0 when it is provided in detectpar

Other
o sim.capthist() for telemetry data updates perimeter (MCP) of
simulated points
o new score.test() arguments control numerical gradient
calculations by fdHess
o secr.fit() with method = "none" now reports the log likelihood
at the parameter values in "start", as well as asymptotic
variances
o secr.fit() now returns a density estimate when CL = FALSE and all detection
parameters are fixed
o secr.fit() now accepts character values for the 'capthist' and
'mask' arguments; objects thus named are retrieved from the workspace.
o read.traps() and read.capthist() now return an error if any
coordinate is not numeric
o components of an secrlist may now be extracted for AIC() with '['
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand

Return to software problems/news

Who is online

Users browsing this forum: No registered users and 2 guests