dsdive.obs.sample.offsets: Sample t0 and tf offsets from full conditional posteriors

In jmhewitt/dsdive: Discrete Space Models for Dive Trajectories of Animals

Description

Sampler uses a piecewise quadratic polynomial approximation to the log of the full conditional posterior. Note that this sampler will only sample one offset at a time.

Usage

dsdive.obs.sample.offsets(
  dsobs.aligned,
  dsobs.unaligned,
  offset,
  offset.tf,
  t.stages,
  P.raw,
  depth.bins,
  tstep,
  max.width,
  prior.params,
  sample.start,
  debug = FALSE,
  lpapprox = NULL,
  output.lpapprox = FALSE
)

Arguments

dsobs.aligned	dsobs object that has been adjusted to account for the t0 and tf offsets. Note that this should be based on the current value of the offsets, and that this function will update these parameters.
dsobs.unaligned	dsobs object with the raw observations of a dive object
offset	value of current t0 offset
offset.tf	value of current tf offset
t.stages	stage transition times for the dive
P.raw	list of continuous time probability transition matrices, and components.
depth.bins	n x 2 Matrix that defines the depth bins. The first column defines the depth at the center of each depth bin, and the second column defines the half-width of each bin.
tstep	Time between observations in dsobs.unaligned
max.width	The t0 and tf offsets are updated with a piecewise proposal distribution. This parameter controls the maximum width of the intervals for the proposal distribution. This is a tuning parameter that controls the numerical stability of the proposal distribution, which is sampled via inverse CDF techniques.
prior.params	shape1 and shape2 parameters for the scaled and shifted Beta prior distribution for the offset being sampled.
sample.start	TRUE to sample t0 offset; FALSE to sample tf offset.
debug	TRUE to return debugging objects, such as the proposal density and log posterior
lpapprox	(Optional) Object containing the piecewise-quadratic approximation to the log-posterior used to propose model parameters. If NULL, then an approximation will be computed.
output.lpapprox	TRUE to return the approximation used to propose model parameters. # @example examples/dsdive.obs.sample.offsets.R

jmhewitt/dsdive documentation built on May 29, 2020, 5:18 p.m.