Batched numpy array indexing

Question

I have source and destination numpy arrays, say

dest = np.arange(1000)
src = np.random(500)

I want to copy across the src to the dest array overwriting the first 50 numbers, then all the numbers starting at index 100 to 150, then from 200 to 250 and so on till 900 to 950.

To calculate the index I use the following:

index = np.reshape([100*i+np.arange(50) for i in range(10)],-1)

then I just use

dest[index]=src

to copy across the remaining elements (with another src, say src2), i just adust the index

index2 = np.reshape([50+100*i+np.arange(50) for i in range(10)],-1)
dest[index2]=src2

I'm pretty sure there's a more elegant/efficient way of achieving this without having to explicitly build the indexes.

Is there a better way to perform the copy?

Haminaa · Accepted Answer · 2016-12-21 10:33:03Z

1

dest.reshape(-1,50)[::2] = src.reshape(-1,50)

And for src2:

dest.reshape(-1,50)[1::2] = src2.reshape(-1,50)

edited Dec 21, 2016 at 10:33

answered Dec 21, 2016 at 10:16

Haminaa

3982 silver badges8 bronze badges

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

Comments

PidgeyUsedGust · Accepted Answer · 2016-12-21 10:32:55Z

I guess the easiest would be to reshape both into matrices

dest = dest.reshape((10, 100))
src = src.reshape((10, 50))

and use matrix indexing

dest[:,:50] = src

The idea is to create a 100 x 10 matrix of dest and replace the left half with src converted to a 50 x 10 matrix.

Edit

@Haminaa's answer is slightly faster.

%timeit dest.reshape((10, -1))[:,:50] = src.reshape((10, -1))

The slowest run took 11.54 times longer than the fastest. This could mean that an intermediate result is being cached. 1000000 loops, best of 3: 1.55 µs per loop

%timeit dest.reshape(-1,50)[::2] = src.reshape(-1,50)

The slowest run took 12.97 times longer than the fastest. This could mean that an intermediate result is being cached. 1000000 loops, best of 3: 1.38 µs per loop

Collectives™ on Stack Overflow

Batched numpy array indexing

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related