How to create an array of dimension n+1 given a function returning an array of dimension n

Question

With numpy and python3 I have to following problem:

I have a function which returns a 2 dimensional array of integers of fixed size (2x3 in this case). What is the most idiomatic way to run this function n times and stack these together to a 3 dimensional 2x3xn array? What about performance? Something which only does the minimum number of allocations would be nice.

For actual performance, it would make sense to know about the implementation of that function itself. It could probably be vectorized with NumPy ufuncs and such. — Divakar
– Divakar, Commented Nov 19, 2015 at 20:00
Something which only does the minimum number of allocations would be nice. - that's C++-style thinking. Allocation isn't what's going to kill your performance in Python. If you try to run this function n times instead of taking advantage of NumPy vectorized operations, interpreter overhead is going to kill your performance. — user2357112
– user2357112, Commented Nov 19, 2015 at 20:25

Daniel · Accepted Answer · 2015-11-19 19:58:47Z

2

You are probably looking for np.dstack:

>>> import numpy as np
>>> arrs = [np.random.rand(2, 3) for x in range(5)]
>>> np.dstack(arrs).shape
(2, 3, 5)

If you know the final shape you can do something like the following:

>>> out = np.empty((2, 3, 5))
>>> out[..., 0] = np.random.rand(2, 3)

answered Nov 19, 2015 at 19:58

Daniel

19.6k7 gold badges64 silver badges74 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

How to create an array of dimension n+1 given a function returning an array of dimension n

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related