How to create a list of n arrays from a list of tuples, each tuple containing n arrays? (Other than with a for loop)

Question

The issue

I have a list which contains 4 tuples. It is the output of multiprocessing.Pool.map() , but I don't think that's important.

Each tuple contains 3 numpy arrays.

What is a good way to create 3 arrays, ie append (vstack) all the first arrays into 1, all the second into another, etc? Ie create the orange output from the orange arrays, etc, in the screenshot below:

What I have tried

I could of course do a very banal loop, like in the toy example below; it works, but it doesn't seem very elegant. I presume there's a more elegant/ pythonic way?

x = np.random.rand(10,2)

a = ((x,2*x,3*x))
b = a
c = a
d = a
my_list =[a,b,c,d]

num_items = len(my_list[0])
out =[None] * num_items

for i in range(num_items): #3 arrays in each tuple
    out[i] =[]
    for l in my_list: 
        out[i].append( l[i] )
    out[i] = np.vstack(out[i])

Answer 1

my_array = np.array(my_list).swapaxes(0,1) # puts the `out` dimension in front
my_array.shape
Out[]: (3, 4, 10, 2)

if you want to concatenate the first dimension:

 my.array.reshape(3, -1, 2) #-> shape (3, 40, 2)

if you really want a list:

list(my_array) #-> list of 3 arrays of shape (4, 10, 2)

Answer 2

You can try:

np.concatenate(my_list, axis=1)