I defined a function matrix.create() that generates a m x n Matrix and I'll use a loop to generate matrices for every t = 100 timesteps. The result should be one big matrix of dimension m*t x n.
Is there an easy way to do this? Should I concatenate the matrices within the function?
Here's how (one way) to avoid growing the matrix (inefficient: see e.g. here or chapter 2 here) when every matrix is different:
res <- list()
for(t in 1:n){
res[[i]] <- matrix.create(...)
}
mat <- do.call("rbind", res)
Note that growing a list by appending objects (which is basically what we're doing here, although in an R-ish way) does not incur the same penalty as growing vectors, matrices, or data frames.
It might be slightly more efficient to do this:
nr <- 10; nc <- 10 ## or whatever
mat <- matrix(NA_real_, nrow = nr*n, ncol = nc)
ind <- 1
for (t in 1:n) {
mat[ind:(ind+nr),] <- matrix.create(...)
ind <- ind + nr
}
Instead of incrementing a current-index variable, you could also do this by computing the appropriate indices at each step (i.e., ((t-1)*nr + 1):(t*nr))