Picking eigenvalues below a certain value.

I have a problem where I have, lets say 1000 2x2 random matrices distributed with mean 0 and variance 1. I have to find the eigenvalues below 0.3 and use their corresponding eigenvectors to calculate a quantity from it. So far I have the following code for this exact type of problem:

 [V,D] = eig(H); 
        d = diag(D);
        idx = abs(d)<0.3;
        E1 = d(idx);
        E2 = V(:,idx);

The problem that I'm having is that because the matrices are random in some cases there will be no eigenvalues less than 0.3 and so when that gets put into the calculation Matlab just outputs NaN or Inf. Is there a way to ignore those and only use the matrices where they have eigenvalues less than 0.3?

[John D'Errico]

Be careful, as in some cases, there might be 0, 1, or 2 eigenvalues that satisfy the requirement.

Answer 1

[the cyclist]

You could insert subsequent code inside a statement like

if any(idx)
   ...
end

To guard against there being multiple eigenvalues meeting the criterion, you could change that condition to

if sum(idx)==1
   ...
end

Of course, the correct approach will depend on how you want to handle these different cases.

You could use a switch/case construct to do different things for different cases.