theta2mcmclist: Convert theta to a mcmc List

In ggdmc: Cognitive Models

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)

ggdmc documentation built on May 2, 2019, 9:59 a.m.