Numpy Equivalent of Logic Operation on Multidimensional Arrays in MATLAB

Question

I have the following logic operation coded in MATLAB, where [A, B, C, and D] are all 5x3x16 doubles and [a, b, c, and d] are all 240x1 doubles. I am trying to implement the same logic operation in python using numpy.

D = zeros(size(A)); 
for i = 1:numel(D)
    flag = ...
        (a == A(i)) & ...
        (b == B(i)) & ...
        (c == C(i));
    D(i) = d(flag);
end

d is a column vector that is already populated with data. a, b, and c are also populated column vectors of equal size. Meshgrid was used to construct A, B, and C into a LxMxN grid of the unique values within a, b, and c. Now I want to use d to populate a LxMxN D with the appropriate values using the boolean expression.

I have tried:

D= np.zeros(np.shape(N))
for i in range(len(D)):

    for j in range(len(D[0])):

        for k in range(len(D[0][0])):    
            flag = np.logical_and(                                  
                (a == A[i][j][k]),              
                (b == B[i][j][k]),        
                (c == C[i][j][k])
                )
            D[i][j][k] = d[flag];

Possible duplicate of [How to perform element wise boolean operations on numpy arrays](https://stackoverflow.com/questions/8632033/how-to-perform-element-wise-boolean-operations-on-numpy-arrays) — Cris Luengo, Mar 21 '19 at 17:28
Welcome to the site. Please make sure to provide (cut down) sample data and target output. This will make it a lot easier to help. — mhhollomon, Mar 21 '19 at 17:45
I'm really not clear on what you want `D` to be at the end of this. Can you provide some example data (e.g. A, B, and C, as 2x2x2 and a, b, c, and d as 3x1) with what you'd like/expect the output to be? — aganders3, Mar 21 '19 at 22:51
@aganders3 It is hard to give an example but I will try to explain. d is a column vector that is already populated with data. a, b, and c are also populated column vectors of equal size. Meshgrid was used to construct A, B, and C into a LxMxN grid of the unique values within a, b, and c. Now I want to use d to populate a LxMxN D with the appropriate values using the boolean expression. — MGoforth, Mar 22 '19 at 21:14

score 1 · Answer 1 · answered Mar 21 '19 at 17:25

1

The syntax will be a little messier, but you can use the np.logical_* functions to do this.

answered Mar 21 '19 at 17:25

aganders3

5,433
23
27

Right. I thought that this function would do the trick. I have been struggling with the syntax. – MGoforth Mar 21 '19 at 17:28
Why not `&` and `|`? – Cris Luengo Mar 21 '19 at 17:29
@CrisLuengo I don't think it works the same with multidimensional np arrays. I get a ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() – MGoforth Mar 21 '19 at 17:37
@MGoforth: [here](https://stackoverflow.com/a/8632049/7328782) it says that should work. – Cris Luengo Mar 21 '19 at 17:43
@MGoforth: `a=np.random.rand(10,12,6); b=(a>0.4)&(a<0.6)` works for me. – Cris Luengo Mar 21 '19 at 17:46
@CrisLuengo The only difference I see is that they already have a filled array to begin with to construct their mask. I have an empty array that I want to fill when all the element wise conditions are true. – MGoforth Mar 21 '19 at 17:49
@CrisLuengo Thanks. I am going to work on that for a bit. – MGoforth Mar 21 '19 at 17:53

Numpy Equivalent of Logic Operation on Multidimensional Arrays in MATLAB

1 Answers1