simulate_burnin_single: Find the burn-in step
In isci1102/ERPM:

Description Usage Arguments Author(s) Examples

View source: R/functions_burninthining.R

Function used to find appropriate burn-in step.

1	simulate_burnin_single(partition, theta, nodes, effects, objects, num.steps, neighborhood, sizes.allowed, sizes.simulated)

`partition`	Observed partition.
`theta`	Initial model parameters
`nodes`	Data frame containing the nodes.
`effects`	Effects or sufficient statistics. A list with a vector "names" and a vector "objects".
`objects`	Objects used for statistics calculation. A list with a vector "name" and a vector "object".
`num.steps`	Number of samples wanted.
`neighborhood`	Way of choosing partitions: probability vector (probability actors swap, probability merge/division, probability single actor move).
`sizes.allowed`	Vector of group sizes allowed in sampling.
`sizes.simulated`	Vector of group sizes allowed in the Markov chain but not necessraily sampled.

Marion Hoffman

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (partition, theta, nodes, effects, objects, num.steps,
    neighborhood, sizes.allowed, sizes.simulated)
{
    num.effects <- length(effects$names)
    print("Neighborhood: ")
    print(neighborhood)
    chain <- draw_Metropolis_single(theta, partition, nodes,
        effects, objects, 1, 1, num.steps, neighborhood, sizes.allowed,
        sizes.simulated, return.all.partitions = F)
    allmeans <- matrix(0, nrow(chain$draws), num.effects)
    allmeans[1, ] <- chain$draws[1, ]
    for (d in 2:nrow(chain$draws)) {
        allmeans[d, ] <- colMeans(chain$draws[1:d, ])
    }
    smoothedmeans <- allmeans
    for (eff in 1:num.effects) {
        lo <- loess(y ~ x, data.frame(x = 1:nrow(chain$draws),
            y = allmeans[, eff]))
        smoothedmeans[!is.na(smoothedmeans[, eff]), eff] <- lo$fitted
    }
    return(list(draws = chain$draws, moving.means = allmeans,
        moving.means.smoothed = smoothedmeans))
  }