Merge numpy arrays returned from loop

Question

I have a loop that generates numpy arrays:

for x in range(0, 1000):
   myArray = myFunction(x)

The returned array is always one dimensional. I want to combine all the arrays into one array (also one dimensional.

I tried the following, but it failed

allArrays = []
for x in range(0, 1000):
   myArray = myFunction(x)
   allArrays += myArray

The error is ValueError: operands could not be broadcast together with shapes (0) (9095). How can I get that to work?

For instance these two arrays:

[ 234 342 234 5454 34 6]
[ 23 2 1 4 55 34]

Shall be merge into this array:

[ 234 342 234 5454 34 6 23 2 1 4 55 34 ]

maybe concatenate((a,b),1) or hstack((a,b)) See also stackoverflow.com/questions/18404077/… — Tengis
– Tengis, Commented Mar 26, 2014 at 13:53

jmetz · Accepted Answer · 2014-03-26 15:13:37Z

12

You probably mean

allArrays = np.array([])
for x in range(0, 1000):
    myArray = myFunction(x)
    allArrays = np.concatenate([allArrays, myArray])

A more concise approach (see wims answer) is to use a list comprehension,

allArrays = np.concatenate([myFunction(x) for x in range])

edited Mar 26, 2014 at 15:13

answered Mar 26, 2014 at 13:51

jmetz

12.9k3 gold badges32 silver badges41 bronze badges

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

4 Comments

ustroetz Over a year ago

If I use your first example. I get the error TypeError: only length-1 arrays can be converted to Python scalars

jmetz Over a year ago

@ustroetz: Thanks for the catch, forgot the brackets as concatenate takes the arrays to be concatenated as a list.

Christian Vincenzo Traina Over a year ago

It doesn't work, it says

all the input arrays must have same number of dimensions, but the array at index 0 has 1 dimension(s) and the array at index 1 has 3 dimension(s)

jmetz Over a year ago

If you check the OP, this question and answer are for the case of 1-dimensional output arrays. It's clear from your comment that the output arrays you have are not always 1-dimensional (notice how the error message references a 3-dimensional array).

Никита Черкес · Accepted Answer · 2018-03-28 09:14:55Z

11

You should know the shape of returned array. Suppose, myArray.shape = (2, 4) Then

allArrays = np.empty((0, 4))
for x in range(0, 1000):
    myArray = myFunction(x)
    allArrays = np.append(allArrays, myArray, axis = 0)

answered Mar 28, 2018 at 9:14

Никита Черкес

1291 silver badge4 bronze badges

2 Comments

PirateApp Over a year ago

doesnt this have the first row as the empty values?

Никита Черкес Over a year ago

no, because we create array with 0 rows and 4 columns: array([], shape=(0, 4), dtype=float64)

wim · Accepted Answer · 2023-04-07 16:51:56Z

10

It sounds like you want to use np.concatenate:

arrays = [myFunction(x) for x in range(1000)]
allArrays = np.concatenate(arrays)

edited Apr 7, 2023 at 16:51

answered Mar 26, 2014 at 13:52

wim

368k114 gold badges681 silver badges816 bronze badges

Collectives™ on Stack Overflow

Merge numpy arrays returned from loop

3 Answers 3

4 Comments

2 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

4 Comments

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related