data.init.2: Initialization

In mpoll/scale: The Scaleable Langevin Exact Algorithm -- Reference Implementation

Usage

data.init.2(beta.eval = beta.star, datum.idx.start = 1, length.idxs = dsz)

Arguments

beta.eval
datum.idx.start
length.idxs

Examples

##---- 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 (beta.eval = beta.star, datum.idx.start = 1, length.idxs = dsz) 
{
    call.data <- data.subset(datum.idx.start, length.idxs)
    subset.design <- call.data$subset.design
    subset.data <- call.data$subset.data
    subset.zs <- as.numeric(apply(subset.design, 1, function(x) exp(sum(x * 
        beta.eval))))
    subset.grad.log.pi <- sapply(seq_len(length.idxs), function(i) diag(n.sigma) * 
        subset.design[i, ] * (subset.data[i] - subset.zs[i]/(1 + 
        subset.zs[i])))
    subset.lap.log.pi <- sapply(seq_len(length.idxs), function(i) -(diag(n.sigma) * 
        subset.design[i, ])^2 * subset.zs[i]/(1 + subset.zs[i])^2)
    grad.log.pi <- as.numeric(apply(subset.grad.log.pi, 1, sum))
    lap.log.pi <- as.numeric(apply(subset.lap.log.pi, 1, sum))
    alpha.cent <<- matrix(grad.log.pi, nrow = 1)
    alpha.cent.sq <<- (alpha.cent) %*% t(alpha.cent)
    alpha.p.cent <<- sum(lap.log.pi)
    phi.cent <<- (alpha.cent.sq + alpha.p.cent)/2
    list(alpha.cent = alpha.cent, alpha.cent.sq = alpha.cent.sq, 
        alpha.p.cent = alpha.p.cent, phi.cent = phi.cent)
  }

mpoll/scale documentation built on Dec. 9, 2019, 7:15 a.m.