Pairwise combination of arbitrary set of lists

Question

I am looking for a generalized approach that solves the following problem: Given an arbitrary number of one- or multidimensional lists (NumPy arrays or the like) return their pairwise combinations.

The following tries to give an example for three one-dimensional lists a , b and c that result in d :

a = np.linspace(0, 1, 1).astype(int)
b = np.linspace(1, 2, 2).astype(int)
c = np.linspace(2, 4, 3).astype(int)

e = np.array([
    (a_, b_, c_)
    for a_ in a
    for b_ in b
    for c_ in c
])

When executed the variables are set as follows:

# a
[0]

# b
[1 2]

# c
[2 3 4]

# d
[[0 1 2]
 [0 1 3]
 [0 1 4]
 [0 2 2]
 [0 2 3]
 [0 2 4]]

What would be a good way for a generalized approach? I'm ideally looking for a function that takes an iterable whose elements define the one- or multidimensional lists like so:

def pairwise_combinations(iterable):
    # Insert magic here

Answer 1

itertools.product got your back: https://docs.python.org/3/library/itertools.html#itertools.product

In the given example it could be used like this:

d = np.array(list(itertools.product(a, b, c)))

Answer 2

I believe the itertools.product function will do exactly what you want. It takes iterables as arguments, and forms tuples of all-to-all, eg:

> a = [0]
> b = [1, 2]
> c = [2, 3, 4]
> print(list(product(a,b,c)))
[(0, 1, 2), (0, 1, 3), (0, 1, 4), (0, 2, 2), (0, 2, 3), (0, 2, 4)]

Details on that set of tools can be found here .