Is Adding Booleans Correct for Counting True Values in a Vector

Is it conceptually correct to sum a vector of booleans? From a mathematical point of view, I would argue it's not: True + True != 2. But it's quite practical to do so still! Example using the vectorised Python library numpy:

In [1]: X = rand(10)

In [2]: large = X>0.6

In [3]: large.dtype
Out[3]: dtype('bool')

In [4]: large.sum()
Out[4]: 7

I don't like it, but it's very practical. Is this a good practice?

Update: the aim is to count the number of true values in a vector.\

Update 2013-02-18: I just discovered the numpy function count_nonzero does exactly what I need in a proper way. That means that as far as Python is concerned, there is no need to use the "dirty" way.

Best Answer

I would say it is not semantically proper to add True and True to get two. It may work, but it seems to rely on an implementation detail.

Python defines a few things as False, such as:

""
0
0.00
None, 
[], 
(),
.__bool__() evaluates to False, 
etc...

and everything else is True.

The + operator in python, when used on a bool will first convert it to int.

>>> int(True)
1
>>> int(False)
0
>>> True + True
2

If you choose to do so in Python, make sure you comment your code accordingly.