[Tutor] Python question

[Peter Otten]

On 18/04/2022 11:50, Eleonora Panini wrote:
> <https://stackoverflow.com/posts/71904558/timeline>
>
> I need help to do the following steps in Phyton:
>
> -my starting array is: array = np.random.rand(1024x11700)
>
> -I have to obtain an array 1024x23400 with the following characteristics:
> 25 columns(1024x25) of data and 25 columns (1024x25) of zeros
>
>   So I thought to do a loop every 25 columns(1024x25), insert an array of
> zeros (1024x25) alternatevely every 25 columns, so I obtain an array
> 1024x23400..
>
> but I don't know how to do it in Phyton.. Can you help me? Thank you

Hello Eleonora,

in case you are still not working ;) on the problem here's some
spoon-feeding. My approach differs from what you sketched out above in
that I create a result matrix (i. e. your 1024x23400 beast) containing
only zeros and then copy parts of the original data (the 1024x25 chunks)
into that result matrix.

Notes:

- For aesthetic reasons I swapped rows and columns
- Numpy experts might do this differently

import numpy


def blowup(a, chunksize):
     assert len(a.shape) == 2
     if a.shape[0] % chunksize:
         raise ValueError(
             f"Number of matrix rows ({a.shape[0]}) "
             f"must be a multiple of chunk size ({chunksize})."
         )

     # create a matrix of zeros with twice as many rows as 'a'
     b = numpy.zeros((2 * a.shape[0], a.shape[1]))

     # lazily split 'a' into sub-matrices with chunksize rows
     chunks = (
         a[start:  start + chunksize]
         for start in range(0, a.shape[0], chunksize)
     )

     # iterate over the sub-matrices and replace zeros in 'b' with those
     # sub-matrices.
     delta = 2 * chunksize
     start = 0
     for chunk in chunks:
         b[start:start+chunksize] = chunk
         start += delta

     return b


# sample 6x3 matrix
a = numpy.arange(6 * 3, dtype=float).reshape((6, 3)) + 1
print(a)

# create 12x3 matrix with every two data rows followed by two rows of
# zeroes
b = blowup(a, 2)
print(b)

# 12x3 matrix with every three data rows followed by three rows of
# zeroes
print(blowup(a, 3))

# 3x3 matrix cannot be broken into chunks of two without rest
# --> ValueError
print(blowup(a[:3], 2))