Friday, January 27, 2012

Replicating NetLogo Fire model

While preparing for the new semester I have started reimplementing standard NetLogo examples in R. The first is Fire model.
The simulation in R is presented here:

# Forest matrix trees encoding:
# 3 – green, 2 – burning, 1 – burnt, 0 – no tree

simulate <- function (forest.density = 0.6, size = 251) {
      forest <- matrix(3 * rbinom(size ^ 2, 1, forest.density),
                     nrow = size)
      forest[1,] <- 2

      while (sum(forest == 2) > 0) {
            list.red <- which(forest == 2)
# find spots touching burning trees
            ignite.x <- 1 + floor((list.red - 1) / size)
            ignite.x <- rep(ignite.x, each = 4) +
                    rep(c(1, -1, 0,0), length(ignite.x))
            ignite.y <- 1 + ((list.red - 1) %% size)
            ignite.y <- rep(ignite.y, each = 4) +
                    rep(c(00, 1,-1), length(ignite.y))
            include <- (ignite.x > 0) & (ignite.x <= size) &
                   (ignite.y > 0) & (ignite.y <= size)
            ignite.n <- (ignite.x[include] - 1) * size +
                    ignite.y[include]
# ignite green trees
            is.green <- (forest[ignite.n] == 3)
            forest[ignite.n[is.green]] <- 2
# trees burn for one period
            forest[list.red] <- 1
      }
      return(1 - sum(forest[-1,] == 3) / sum(forest[-1,] > 0))
}

It is interesting to notice that it is quite compact. The only exception is handling of finding neighbors which is much easier in NetLogo.

The only thing left was to check that the results produced by NetLogo and R implementations are identical. I have simulated both models 1000 times for forest density equal to 60%. Here is the plot comparing percent burned density estimates and Kolmogorov-Smirnov test p-value.


It can be concluded that for the parametrization I have selected models produce similar results.

No comments:

Post a Comment