How to create an array of matrices in python?

Question

In MATLAB, I do something like this:

for i = 1:n
    a{i} = f(i)
end

where f(i) is a function that returns a random matrix of fixed size.

In python, I am working with numpy but I do not understand how to do this.

import numpy as np
a = np.array([])
for i in range(0, n):
   # a.insert(i, f(i)) and does not work
   # a[i] = f(i)  and does not work

Do you know the shape of the array returned by f in advance? — Will Vousden
– Will Vousden, Commented May 1, 2015 at 16:34

Will Vousden · Accepted Answer · 2015-06-15 08:07:02Z

2

If you want a rank-3 Numpy array, and you know the shape of f(i) in advance, you can pre-allocate the array:
```
a = np.zeros((n,) + shape)
for i in range(n):
    a[i] = f(i)
```
If you just want a list (rather than a Numpy array) of matrices, use a list comprehension:
```
a = [f(i) for i in range(n)]
```
Another way to get a Numpy array is to convert from the list comprehension above:
```
a = np.array([f(i) for i in range(n)])
```
However, this will be less efficient than #1 because the results of f(i) are first buffered in a dynamically-sized list before the Numpy array is built.

answered May 1, 2015 at 16:38

Will Vousden

33.5k9 gold badges89 silver badges97 bronze badges

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

Comments

Daniel · Accepted Answer · 2015-09-15 09:38:20Z

2

The best equivalent of a matlab cell array here is a list in python:

a = []
for i in range(n):
    a.append(f(i))

answered May 1, 2015 at 16:35

Daniel

42.9k4 gold badges57 silver badges82 bronze badges

left-over from the question.