How can I select three random sample of a matrix with minimum separation between samples?

Hi everyone,

I have the following problem: I have a matrix A(20x20) per example and I need select from this matrix 3 random samples that have to be different and with a minimum separation (distance) between samples. I mean that between samples chosen randomly, would exist a minimum separation of 4 samples, per example the sum of 3 samples to the right in the same row and 1 sample up in the column.

There exists a way to obtain this?

I have been trying to get this using the following, but if I repeat it 3 times, exists the probability to get the same sample. Also I don't know how to get the restriction of the minimum separation.

random_sample1 = A(randi(numel(A)))

Thank you very much in advance for your help.

J.F.

[James Tursa]

You may need to do this either sequentially or with a rejection method to satisfy your distance requirement.

[Javier Fuster]

Yes, I think so. I was wondering if there was a method more robust than start to reject putting so many 'if'.

Also, thank you for your response.

Answer 1

[James Tursa]

A rejection method is straightforward. Simply make your selections in a forever loop until you get something that works. The advisibility of this approach depends on the probability of getting a rejection. E.g., rejection method code for generating one sample set

n = numel(A);
nsamples = 3;
dmin = 4;
while( true )
    sample_indexes = randperm(n,nsamples);
    [row,col] = ind2sub(size(A),sample_indexes);
    d = abs(row - row') + abs(col-col');
    d(1:nsamples+1:end) = inf;
    if( all(d(:)>=dmin) )
        break;
    end
end
samples = A(sample_indexes);

This assumes there is no wrap-around for row & column comparisons.

[Javier Fuster]

Thank you very much! It works perfectly :)

I am very grateful!