[R] Spatial Voter Model

[Tamas Papp]

On Fri, Apr 16, 2004 at 04:35:15PM -0400, Martin Henry H. Stevens wrote:

>   # select random rows where number of reps = number of cells in arena
>   # and same for columns
> ri <- sample(1:L, replacements, replace=TRUE)
> rj <- sample(1:L, replacements, replace=TRUE)
> 
> #implement a voter model where each cell takes on the value of a 
> randomly selected individual around it.
> 
> for(i in 1:replacements) Lat[ri[i],rj[i]] <- 
> sample(c(Lat[loc[ri[i]],loc[rj[i]+1]],
>                                                       
> Lat[loc[ri[i]+1],loc[rj[i]+2]],
>                                                       
> Lat[loc[ri[i]+2],loc[rj[i]+1]],
>                                                       
> Lat[loc[ri[i]+1],loc[rj[i]]]), 1)

I don't think that you need a loop for this.  Look at the help page of
the [ operatot (type ?"[" at the prompt).  You need something like this:

## this gives you a row & column index matrix (elements to replace)
r <- matrix(sample(1:L, 2*replacements, replact=TRUE), replacements, 2)

## offset matrix - you only need to create this once
roff <- matrix(c(0,1,
		 1,2,
		 2,1,
		 1,1), 4, 2, byrow=TRUE)

## make random neighbor offset selection
n <- sample(1:nrow(roff), replacements, replace=TRUE)

## row & column offsets
roffij <- t(sapply(n, function (i) { roff[i,1:2] }))

## replace elements
Lat[r] <- Lat[r + roffij]

I haven't tested this, please check everything, and look at the help
pages for further information.

> # Calculate Simpon's Diversity (Nei genetic heterozygozity)
> 1 - sum( (table(Lat0)/sum(table(Lat0)))^2)
> 1 - sum( (table(Lat)/sum(table(Lat)))^2)

You might be able to speed this up by writing some C code, and calling
it in R.  It is only worth it if you have large matrices, and use this
kind of thing really often (and you can program C).  Maybe somebody
has a better suggestion.

Best,

Tamas

-- 
Tamás K. Papp
E-mail: tpapp at axelero.hu
Please try to send only (latin-2) plain text, not HTML or other garbage.