Find the minimum non zero value in a matrix

Question

I'm trying to find a 2d array that represents the minimum values of the 3rd dimension in a 3d array., eg

a = floor(rand(10,10,3).*100); % representative structure
b = min(a,[],3); % this finds the minimum but also includes 0 

I tried using:

min(a(a>0),3) 

but that isn't correct? I guess I could sort the third dimension of a and then find the minimum within 1:depth-1 - but that doesn't seem the most efficient way?

Any thoughts?

Answer 1

The problem is that a(a>0) returns a linear array, so you'll end up with one minimum, as opposed to a 2D array with minima.

The safest way to take the minimum of non-zero values is to mask them with Inf , so that the zeros do not interfere with the calculation of the minimum.

tmp = a;
tmp(tmp==0) = Inf;

b = min(tmp,[],3);

Answer 2

One possibility would be to simply make all the zero values very big.

For example, if you know that no elements would ever be larger than 1000 you could use

b = min(a+1000*(a==0),[],3)

Answer 3

simply assign those points infinity where the value is zero so always the min answer will not count zero ones..... like a(a==0)=inf; %then count the min one minelement=min(a);

Answer 4

remove zero elements from matrix like this:

    a = [10 2 0 4 5; 156 1.7 45 23 0 ];
    a(a == 0) = NaN;% not a number
    min(a(:))
    >> ans = 1.7