Computing the capacity of rewritable memories

We propose an algorithm for computing the capacity of discrete rewritable storage devices subject to a constraint on the maximal number of rewrite operations. The linchpin is that-although the number of writing strategies is exponential in the maximal number of allowed rewrites-linear functionals of the probabilities they induce on the output space can be efficiently maximized using Dynamic Programming.