numpy: broadcasting ndarray of booleans

Question

I have a situation in which I have an ndarray X of floats, let's say 100x10, and I want to look at some conditions on the first column and create a boolean ndarray B of shape 100x1. Then I want to use B as an index into X to pull out values where a True is located. However for each True in BI want to pull out the entire row of X. I thought this would work automatically, as B would be broadcast to a 100x10 shape. However it doesn't seem to work this way. Here's an example using 2x2 and 2x1 ndarrays.

a = np.array([True, False])
a.shape = (2,1)
b = np.array([1, 2, 3, 4])
b.shape = (2,2)
print(a)
print(b)
print(b[a])

This prints

[[True]
 [False]]

[[ 1 2 ]
 [ 3 4 ]]

 [1]

I expected it to print [1 2] . Why doesn't the broadcasting work the way I expect?

Answer 1

The rules for so-called "fancing indexing" are detailed here . In particular, when the index, obj , is a NumPy array of dtype bool , x[obj]

... is always equivalent to (but faster than) x[obj.nonzero()] where, as described above, obj.nonzero() returns a tuple (of length obj.ndim) of integer index arrays showing the True elements of obj.

Since,

In [4]: a.nonzero()
Out[4]: (array([0]), array([0]))

b[a] is equivalent to b[a.nonzero()] which is

In [6]: b[(np.array([0]), np.array([0]))]
Out[6]: array([1])
In [7]: b[a]
Out[7]: array([1])

---

If you want to use a boolean array a to select rows of b , then, as Joran Beasley states, just keep a as a 1-dimensional boolean array:

import numpy as np

a = np.array([True, False])
b = np.array([1, 2, 3, 4])
b.shape = (2,2)
print(b[a])    
# [[1 2]]