[new Model] Binary Correlated Process

[discoleo]

Model: Binary Correlated Process

Create a new page for a "binary"-correlated process.

R Code

### Binary Linear Correlation
# - but Non-Linearly, Non-Monotonically Coupled Process;
rgrid.correl = function(dim, pChange=1/3, type = c("Constant", "Bernoulli")) {
    type = match.arg(type);
    nr = dim[1]; nc = dim[2];
    m = matrix(FALSE, nrow=nr, ncol=nc);
    if(nr == 0 || nc == 0) return(m);
    m[,1] = FALSE;
    if(nc == 1) return(m);
    # Subsequent Columns:
    if(type == "Constant") {
        nCh = round(nr * pChange);
        ids = seq(1, nr);
        for(i in seq(2, nc)) {
            tmp = rep(FALSE, nr);
            id  = sample(ids, nCh);
            tmp[id] = ! tmp[id];
            m[, i]  = tmp;
        }
    } else {
        for(i in seq(2, nc)) {
            tmp = rep(FALSE, nr);
            ids = as.logical(rbinom(nr, 1, pChange));
            tmp[ids] = ! tmp[id];
            m[, i] = tmp;
        }
    }
    return(m);
}

Specialized Function: as.grid.correl

    as.grid.correl = function(x, m.cor, p, val=-1) {
        nr = nrow(m.cor); nc = ncol(m.cor);
        if(length(x) != nr) stop("Invalid dimensions of the carry-forward matrix!");
        m = matrix(0, nrow=nr, ncol=nc);
        if(nr == 0 || nc == 0) return(m);
        m[x > p, 1] = 1;
        as.grid0 = function(m) {
            m[m == 0] = val;
            m[m > 0]  = 0;
            return(m);
        }
        if(nc == 1) {
            return(as.grid0(m));
        }
        # Subsequent Columns:
        tmp = m[, 1];
        for(i in seq(2, nc)) {
            tmp[m.cor[, i]] = 1 - tmp[m.cor[, i]];
            m[, i] = tmp;
        }
        return(as.grid0(m));
    }

Example

### Example
d = c(200, 50)
x = runif(d[1])
pChange = 1/3
m.cor = rgrid.correl(d, pChange = pChange)
p = 0.53
m = as.grid.correl(x, m.cor, p=p)
m = flood.all(m)
plot.rs(m)