CODEWITHSUNDEEP
pythonnumpy
Startec
Startec

Reputation: 13206

Understanding the slicing syntax in Numpy?

I have an x dimensional matrix in Numpy. For this example, I will use a 2x2 array.

np.array([[2, 2], [3,3]])

How would I alternate adding a row and column of some value so the result would look like:

array([[2, x, 3, x],
       [x, x, x, x].
       [2, x, 3, x],
       [x, x, x, x]])

This answer gives a helpful start by saying to set rows in a correctly-sized destination matrix b from a matrix a like so a[::2] = b but what does the ::2 do in the slicing syntax and how can I make it work on columns?

In short what do the x y and z parameters do in the following: a[x:y:z]?

Upvotes: 0

Views: 141

Answers (1)

Tacratis
Tacratis

Reputation: 1055

If I understand what you want correctly, this should work:

import numpy as np
a = np.array([[2,2],[3,3]])
b = np.zeros((len(a)*2,len(a)*2))
b[::2,::2]=a

This 'inserts' the values from your array (here called a) into every 2nd row and every 2nd column

Edit: based on your recent edits, I hope this addition will help:

x:y:z means you start from element x and go all the way to y (not including y itself) using z as a stride (e.g. 2, so every 2 elements, so x, x+2, x+4 etc up to x+2n that is the closest to y possible) so ::z would mean ALL elements with stride z (or ::2 for every 2nd element, starting from 0)

You do that for each 'dimension' of your array, so for 2d you'd have [::z1,::z2] for going through your entire data, striding z1 on the rows and z2 on the columns.

If that is still unclear, please specify what is not clear in a comment.

One final clarification - when you type only : you implicitly tell python 0:len(array) and the same holds for ::z which implies 0:len(array):z. and if you just type :: it appears to imply the same as : (though I haven't delved deep into this specific example)

Upvotes: 1

Related Questions