dependent nested for loops in R

z = NULL
p = 0.25
x = rnorm(100)
z[1] = p*x[1] + (1-p)*5
for(i in 2:100)
{
   z[i] = p*x[i]+(1-p)*z[i-1]
}

That's the same as

p <- 0.25
x <- rnorm(100)
z <- stats::filter(p*x, 1-p, init = 5, method="recursive")

This leaves z as a time series; if that causes trouble, follow it with

z <- as.numeric(z)

Duncan Murdoch