Search code examples
pythonarraysnumpyunpackargument-unpacking

Maximum of multiple numpy arrays

We can compute the element-wise maximum of 3 numpy arrays:

import numpy as np
A = np.arange(20).reshape((4, 5))       # any 4x5 array
B = np.maximum(A, A+7, A+2)     # working

But why doesn't np.maximum accept multiple arrays from an "unpacking"?

L = [np.roll(A, k, axis=0) for k in range(4)]     # 4 arrays: A shifted with different k
np.maximum(*L)

Error:

ValueError: invalid number of arguments

After all, L is a Python list of Numpy array objects, so *L should unpack it for the np.maximum function call. Why doesn't it work?

PS: I also tried with L = (...) (which gives a generator) or L = tuple(...), but we have the same error.

Solution

  • As pointed out in a comment, reduce is the solution here:

    np.maximum.reduce([np.roll(A, k, axis=0) for k in range(4)])