dsdive.obs.sampleparams: Sample pi and lambda parameters for a single dive stage

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

Description

Sampler uses a Gaussian approximation to the full conditional posterior.

Usage

dsdive.obs.sampleparams(
  dsobs.list,
  t.stages.list,
  P.raw,
  s0,
  depth.bins,
  beta,
  lambda,
  lambda.priors.list,
  beta.priors.list,
  tstep,
  gapprox = NULL,
  output.gapprox = FALSE,
  delta
)

Arguments

dsobs.list	list of dsobs objects, which describe the observation times and depths of a collection of dives
t.stages.list	list of initial stage transition times for dives observed in dsobs.list
P.raw	list of continuous time probability transition matrices, and components.
s0	stage for which model parameters should be sampled
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.
beta	directional preference model parameters. See dsdive.tx.params for more details.
lambda	diving rate model parameters. See dsdive.tx.params for more details.
lambda.priors.list	list of parameters for the prior distributions for the diving rate parameters. See dsdive.gibbs.obs for more information.
beta.priors.list	list of parameters for the prior distributions for the directional preference parameters. See dsdive.gibbs.obs for more information. Note that these priors are specified as "pi" priors vs. priors for "beta".
tstep	Time between observations in dsobs.unaligned
gapprox	(Optional) gaussapprox object containing the Gaussian approximation used to propose model parameters. If NULL, then a Gaussian approximation will be computed.
output.gapprox	TRUE to return the Gaussian approximation used to propose model parameters.
delta	If delta>0, then the probability transition matrices computed will use a transition matrix whose generator is perturbed to allow much faster computation. See dsdive.obstx.matrix for more details.

Examples

data('dive.sim')
attach(dive.sim)
attach(dive.sim$params)

t.stages = sim$times[c(FALSE,diff(sim$stages)==1)]
  
tstep = diff(sim.obs$times[1:2])

obstx.mat = lapply(1:3, function(s) {
  dsdive.obstx.matrix(depth.bins = depth.bins, beta = beta, 
                      lambda = lambda, s0 = s, tstep = tstep, 
                      include.raw = TRUE, delta = 1e-10)
})

lambda.priors = list(
  c(4, 2 + 2/3), c(4, 13 + 1/3), c(4, 5)
)

beta.priors = list(
  c(31.5, 3.5), c(3.5, 31.5)
)

x = dsdive.obs.sampleparams(dsobs.list = list(sim.obs), 
                            t.stages.list = list(t.stages), P.raw = obstx.mat, 
                            s0 = 1, depth.bins = depth.bins, beta = beta, 
                            lambda = lambda, lambda.priors.list = lambda.priors, 
                            beta.priors.list = beta.priors, 
                            tstep = diff(sim.obs$times[1:2]), delta = 1e-10)

detach(dive.sim$params)
detach(dive.sim)

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