R/plotClusterSize.R

In dclone: Data Cloning and MCMC Tools for Maximum Likelihood Methods

plotClusterSize <-
function(n, size, balancing = c("none","load","size","both"), 
plot = TRUE, col = NA, xlim = NULL, ylim = NULL, main, ...)
{
    clusterSplitLB <- function(cl, seq, size = 1) {
        tmp1 <- tmp2 <- matrix(NA, length(cl), length(seq))
        tmp1[1,1] <- size[1]
        tmp2[1,1] <- seq[1]
        for (i in 2:length(seq)) {
            rs <- rowSums(tmp1, na.rm=TRUE)
            tmp1[which(rs == min(rs))[1],i] <- size[i]
            tmp2[which(rs == min(rs))[1],i] <- seq[i]
        }
        lapply(1:length(cl), function(i) tmp2[i,!is.na(tmp2[i,])])
    }
    if (n == 1)
        stop("'n' > 1 is needed")
    m <- length(size)
    seq <- 1:m
    balancing <- match.arg(balancing)
    cl <- structure(c(1:n), class="cluster")
    x <- switch(balancing,
        "none" = clusterSplit(cl, seq),
        "load" = clusterSplitLB(cl, seq, size),
        "size" = clusterSplitSB(cl, seq, size),
        "both" = clusterSplitSB(cl, seq, size))
    s <- switch(balancing,
        "none" = clusterSplit(cl, size),
        "load" = clusterSplitLB(cl, size, size),
        "size" = clusterSplitSB(cl, size, size),
        "both" = clusterSplitSB(cl, size, size))
    x2 <- lapply(s, cumsum)
    x1 <- lapply(1:n, function(i) x2[[i]] - s[[i]])
    offset <- 0.1
    y <- 1:n
    y1 <- y+(0.5-offset)
    y2 <- y-(0.5-offset)
    col <- rep(col, m)[1:m]
    col <- col[unlist(x)]
    coli <- 1
    MAX <- max(sapply(x2, function(z) if (length(z)) max(z) else 0))
    if (plot) {
        if (is.null(xlim))
            xlim <- range(x1,x2)
        if (is.null(ylim))
            ylim <- c(range(y1,y2)[2], range(y1,y2)[1])
        plot.new()
        plot.window(xlim=xlim, ylim=ylim)
        box()
        axis(side=1)
        axis(side=2, at=y, tick=FALSE, las=1)
        if (missing(main))
            main <- switch(balancing,
                "none" = "No Balancing",
                "load" = "Load Balancing",
                "size" = "Size Balancing",
                "both" = "Size and Load Balancing")
        sub <- paste("Max =", round(MAX, max(3, getOption("digits") - 3)))
        title(main=main, xlab="Approximate Processing Time", ylab="Workers",
            sub=sub)
        for (i in 1:n) {
            for (j in 1:length(x[[i]])) {
                tmp <- try(polygon(c(x1[[i]][j], x2[[i]][j], x2[[i]][j], x1[[i]][j]), 
                    c(y1[i], y1[i], y2[i], y2[i]), col=col[coli], ...), silent=TRUE)
                coli <- coli + 1
                if (!inherits(tmp, "try-error"))
                    text(mean(c(x1[[i]][j], x2[[i]][j])), y[i], x[[i]][j])
            }
        }
    }
    invisible(MAX)
}