Add numpy array in blockwise

Question

I have a numpy array generated by some earlier code which needs to be added blockwise. For example, array a has 32 entries and should be added such that the new array b has 4 entries and b[0] has numbers 0-7, b[1] has 8-15 and so on. See example below of how to do it manually.

import numpy as np
a = np.random.rand(32)
b = np.zeros(4)
b[0] = np.sum(a[0:8])
b[1] = np.sum(a[8:16])
b[2] = np.sum(a[16:24])
b[3] = np.sum(a[24:32])

Now I know that I can do the summation using a for loop, but I was hoping for something more fancy, as I am working with rather large arrays. I am aware of numpy's great slicing magic, so I imagine something like this could be used.

mibrahimy · Accepted Answer · 2020-12-01 09:47:49Z

3

I would use something like this:

#import random
import numpy as np
np.random.seed(11) # random.seed doesnt affect numpy computation as pointed out in the comments 

a = np.random.rand(32)
b = np.zeros(4)
a = np.reshape(a, (4,-1))
b = np.sum(a, axis=1)

edited Dec 1, 2020 at 9:47

answered Dec 1, 2020 at 8:24

mibrahimy

7225 silver badges19 bronze badges

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

4 Comments

adir abargil Over a year ago

What is the -1 means?

mibrahimy Over a year ago

One of the dimensions in reshape can be -1. In this case, the value is inferred from the length of the array and the remaining dimensions. Complete doc here numpy.org/doc/stable/reference/generated/…

laolux Over a year ago

This works great, thanks. Only tiny improvement would be to use np.random.seed, as currently your seed does not affect numpy and every run will get different results. Note that this may be version dependent, I recall some libraries directly using random its internal state, but at least recent versions of numpy keep track of their own state, unaffected by other packages.

mibrahimy Over a year ago

Thanks, will update the answer accordingly.

Collectives™ on Stack Overflow

Add numpy array in blockwise

1 Answer 1

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related