theta2mcmclist {ggdmc}

Convert theta to a mcmc List

Description

Extracts the parameter array (ie theta) from posterior samples of a partiipant and convert it to a coda mcmc.list.

Usage

theta2mcmclist(x, start = 1, end = NA, split = FALSE,
  subchain = FALSE, nsubchain = 3, thin = NA)

phi2mcmclist(x, start = 1, end = NA, split = FALSE,
  subchain = FALSE, nsubchain = 3)

Arguments

x

posterior samples

start

start iteration

end

end iteraton

split

whether to divide one MCMC sequence into two sequences.

subchain

boolean swith convert only a subset of chains

nsubchain

indicate the number of chains in the subset

thin

thinning lenght of the posterior samples

Details

phi2mcmclist extracts the phi parameter array, which stores the location and scale parameters at the hyper level.

Examples

## Not run: 
model <- BuildModel(
p.map     = list(a = "RACE", v = c("S", "RACE"), z = "RACE", d = "1",
            sz = "1", sv = "1", t0 = c("S", "RACE"), st0 = "1"),
match.map = list(M = list(gun = "shoot", non = "not")),
factors   = list(S = c("gun", "non"), RACE = c("black", "white")),
constants = c(st0 = 0, d = 0, sz = 0, sv = 0),
responses = c("shoot", "not"),
type      = "rd")

pnames <- GetPNames(model)
npar   <- length(pnames)
pop.mean  <- c(1, 1, 2.5, 2.5, 2.5, 2.5, .50, .50, .4, .4, .4, .4)
pop.scale <- c(.15, .15, 1, 1, 1, 1, .05, .05, .05, .05, .05, .05)
names(pop.mean)  <- pnames
names(pop.scale) <- pnames
pop.prior <- BuildPrior(
  dists = rep("tnorm", npar),
  p1    = pop.mean,
  p2    = pop.scale,
  lower = c(rep(0, 2), rep(-5, 4), rep(0, 6)),
  upper = c(rep(5, 2), rep(7, 4), rep(2, 6)))
p.prior <- BuildPrior(
  dists = rep("tnorm", npar),
  p1    = pop.mean,
  p2    = pop.scale*10,
  lower = c(rep(0, 2), rep(-5, 4), rep(0, 6)),
  upper = c(rep(10, 2), rep(NA, 4), rep(5, 6)))
mu.prior <- BuildPrior(
  dists = rep("tnorm", npar),
  p1    = pop.mean,
  p2    = pop.scale*10,
  lower = c(rep(0,  2), rep(-5, 4), rep(0, 6)),
  upper = c(rep(10, 2), rep(NA, 4), rep(5, 6)))
sigma.prior <- BuildPrior(
  dists = rep("beta", npar),
  p1    = rep(1, npar),
  p2    = rep(1, npar),
  upper = rep(2, npar))
names(sigma.prior) <- GetPNames(model)
priors <- list(pprior=p.prior, location=mu.prior, scale=sigma.prior)
dat    <- simulate(model, nsim = 10, nsub = 10, prior = pop.prior)
dmi    <- BuildDMI(dat, model)
ps     <- attr(dat, "parameters")

fit0 <- StartNewsamples(dmi, priors)
fit  <- run(fit0)

tmp1 <- theta2mcmclist(fit[[1]])
tmp2 <- theta2mcmclist(fit[[2]], start = 10, end = 90)
tmp3 <- theta2mcmclist(fit[[3]], split = TRUE)
tmp4 <- theta2mcmclist(fit[[4]], subchain = TRUE)
tmp5 <- theta2mcmclist(fit[[5]], subchain = TRUE, nsubchain = 4)
tmp6 <- theta2mcmclist(fit[[6]], thin = 2)

## End(Not run)