Efficient way to index a Numpy array with another array

Question

Given a Numpy array x, and an array of integers y, I want to do something equivalent to:

z = np.array(x[i] for i in y)

Is there a Numpy function/method to do this efficiently without converting back to a list?

Jon Clements · Accepted Answer · 2012-12-18 16:47:54Z

3

If y contains indices that are valid for x, then:

z = x[y]

>>> import numpy as np
>>> x = np.arange(100)
>>> y = np.array([1, 27, 36, 98])
>>> x[y]
array([ 1, 27, 36, 98])

answered Dec 18, 2012 at 16:47

Jon Clements

143k34 gold badges254 silver badges288 bronze badges

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

1 Comment

Ah of course - I was trying that but just realized that my y array had converted to float somewhere along the line.