Processing large matrix faster

Hello everyone

I need to procces a large matrix (10.104.485x3) called "M". The first two columns are longitude and latitude and the third column is sea depth. I want to trace the coordinates that have depth bigger than 80m. So I try to make a matrix called "coods" with all the corresponding coordinates. Here follows my code:

%M = readmatrix('Bathymetry_Data.txt');

j=1;

for i=1:10104485;

if M(i,3)<-80

coods(j,1)=M(i,1); coods(j,2)=M(i,2);

j=j+1;

end

end

My problem is that the loop never seem to end. I tried pausing after 2 -3 hours and the "i" is still in the very beginning. My operational system has 8 gb Ram. Can anyone tell if I do something wrong, or If I can do anything to make the loop go faster.

Any help will be deeply appreciated, Thank's a lot in advance

Answer 1

[Stephen23]

Why are you using a loop? Use logical indexing:

M = readmatrix('Bathymetry_Data.txt');
X = M(:,3)<(-80);
coods = M(X,:);

[Manolis Mylonakis]

Thank you so much, it works perfect.

Sorry but I as you can tell I am new to matlab

Answer 2

[Bjorn Gustavsson]

This ought to be done in one go:

coods = M(:,M(:,3)<-80);

In the case you cannot do it in a vectorized way and cannot determine beforehand how big a matrix will be (which makes it impossible to pre-allocate the array), you should try to do the array-growing in blocks instead of element by element, say a couple of 1000 elements at a time and then crop the array down to size at the end.

HTH