Numpy: Set array element to another array

Question

I'm trying to set an element of a Numpy array to be another Numpy array. I'm not sure on how to do this since every time I try I get ValueError: setting an array element with a sequence.

I know this is possible with Python's list since I can append the new array to list and it will work.

This is an example of what I'm trying to do:

import numpy as np

finalArray = np.zeros(3)
finalList = []

a = np.arange(128).reshape(32,4)
b = np.arange(124).reshape(31,4)
c = np.arange(120).reshape(30,4)

# This works
finalList.append(a)
finalList.append(b)
finalList.append(c)

# This doesn't work
finalArray[0] = a
finalArray[1] = b
finalArray[2] = c

Any ideas on how to do this?

"I'm trying to set an element of a Numpy array to be another Numpy array" - bad move. NumPy might not be a good fit for your program's needs. — user2357112
– user2357112, Commented Jun 2, 2016 at 23:10
I was attempting to have variable shapes. For instance in the description above, a, b, and c have different shapes and I was hoping to have a numpy array filled with those. — Xavier Merino
– Xavier Merino, Commented Jun 3, 2016 at 3:22

Community · Accepted Answer · 2017-05-23 12:15:57Z

1

It would work if

finalArray = np.zeros(3, dtype=object)
finalArray[0] = a

Then finalArray takes object points like a list.

But I hesitate suggesting this because this feature is being abused or misused by beginners.

https://stackoverflow.com/a/37597524/901925 - my answer to another question about a deep copy of an array of arrays.

The concatenate answer is the only alternative that makes sense, given the dimensions of your arrays.

A variation on the concatenate is

finalArray = np.zeros((93,4),a.dtype)
finalArray[:32,:] = a
etc

In other words, make finalArray big enough to receive the elements of the arrays, and copy values.

edited May 23, 2017 at 12:15

CommunityBot

11 silver badge

answered Jun 3, 2016 at 0:37

hpaulj

233k14 gold badges260 silver badges392 bronze badges

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

Comments

Andy Hayden · Accepted Answer · 2016-06-02 23:21:54Z

0

I think you're looking for numpy.concatenate:

In [11]: np.concatenate((a, b, c))
Out[11]:
array([[  0,   1,   2,   3],
       [  4,   5,   6,   7],
       [  8,   9,  10,  11],
       [ 12,  13,  14,  15],
       [ 16,  17,  18,  19],
       ...

This creates a single 93 x 4 numpy array:

In [12]: np.concatenate((a, b, c)).shape
Out[12]: (93, 4)

answered Jun 2, 2016 at 23:21

Andy Hayden

378k110 gold badges640 silver badges546 bronze badges

Comments

Algorithmic Canary · Accepted Answer · 2023-09-12 21:25:11Z

0

Numpy arrays are closer to statically typed. They can only contain the types specified on creation. They are more like the arrays of C or Java. While it may be possible to create a numpy array of numpy arrays (where each sub-array is treated as an generic python object), you most likely want to make a proper 2D numpy array.

To create a two dimensional array, feed a two dimensional list into the np.array function. For example,

finalArray = np.array([[0,0,0],[0,0,0],[0,0,0]])

creates a 3x3 two dimensional array filled with zeros.

edited Sep 12, 2023 at 21:25

answered Jun 2, 2016 at 23:11

Algorithmic Canary

7593 silver badges17 bronze badges

1 Comment

Andy Hayden Over a year ago

You can also do this with np.zeros((3, 3), np.int64). That said, I'm not sure how this solves the question (what is x?) Edit: to be fair nor does mine, the question is a little ill-defined!

Collectives™ on Stack Overflow

Numpy: Set array element to another array

3 Answers 3

Comments

Comments

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related