Quick ways to manipulate Numpy array in array

Question

I would like to find a way to quickly manipulate an array of arrays in Numpy like this one, which has a shape of (10,):

[array([0, 1, 3]) ,array([0, 1, 7]), array([2]), array([0, 3]), array([4]),
 array([5]), array([6]) ,array([1, 7]), array([8]), array([9])]

For instance, I'd like to compute the total number of array elements, which is 16 for the array above, but without doing a for loop since in practice my "nested array" will be quite large.

Thanks!

Define "manipulation". If all you want is the length, it would be better to flatten into a single array. Otherwise your only choice s iteration because these arrays are uneven length. — cs95
– cs95, Commented Dec 31, 2018 at 7:58
That's a list of arrays - or object dtype array. Loops, list comprehensions are the normal tools. np.frompyfunc may be useful in some cases, with a modest speed difference. — hpaulj
– hpaulj, Commented Dec 31, 2018 at 8:04
As shown in the answer, 'flattening' the list into one array with concatenate way help. It depends on the manipulation. — hpaulj
– hpaulj, Commented Dec 31, 2018 at 16:57

Jim Todd · Accepted Answer · 2018-12-31 08:25:09Z

5

One way to find the length of the array in your case is to ravel the nested numpy arrays and then find the length as below:

a = [array([0, 1, 3]) ,array([0, 1, 7]), array([2]), array([0, 3]), array([4]),
 array([5]), array([6]) ,array([1, 7]), array([8]), array([9])]

len(np.concatenate(a).ravel())
#Here we expand the numpy arrays and then flatten it to find the length.

Output:

As per my knowledge, ravel has a better timeit performance time in comparison to for loop.

answered Dec 31, 2018 at 8:25

Jim Todd

1,5881 gold badge11 silver badges15 bronze badges

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

1 Comment

hpaulj Over a year ago

I don't think you need the ravel here.

Collectives™ on Stack Overflow

Quick ways to manipulate Numpy array in array

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related