Avoid loops with mapping toolbox "setltln"

Hi all, I'm looking for a clever way to optimize the following chunk of code. The function is to get a grid of lat, lon coordinates using the reference matrix R, and the indices of Z.

However, with such huge dimensions, this nested for loop is painfully slow! How can I avoid for loops in this case?

[Z, R] = arcgridread('biloxi_ms.asc');
tempLAT=zeros(9721,10801);
tempLON=zeros(9721,10801);

for i =1:9721
    for j = 1:10801
    [tempLAT(i,j), tempLON(i,j)] = setltln(Z, R, i, j);
    end
end

Best, Rob

Answer 1

[Sean de Wolski]

You could build a grid and call setltln on all of it. This will consume a bit more memory since you'll need to store ii, jj. A hybrid approach might be to run a loop over smaller meshgrids.

[ii, jj] = meshgrid(1:9271,1:10801)
[tempLAT, tempLON]= setltln(Z, R, ii, ij);

Also, parallel computing could help here if you have the parallel computing toolbox. Simply, turn the outer loop into a parfor.