Comparing boolean numpy arrays

Question

I have two boolean numpy arrays of similar shape like:

a=[[True,True,False,False]]
b=[[True,False,True,False]]

How can I get an array c where 1 indicates that only array a is true, 2 indicates that only array b is true, 0 where both arrays are false and nan where both are true. So in this case the result should be [[nan,1,2,0]]].

Answer 1

You could use np.select:

In [20]: a = np.array([True,True,False,False])

In [21]: b = np.array([True,False,True,False])

In [23]: np.select([a&~b, b&~a, a&b], [1, 2, np.nan], default=0)
Out[23]: array([ nan,   1.,   2.,   0.])

Answer 2

You could use np.where -

np.where(a*b,np.nan,(2*b + a))

Sample run -

In [60]: a
Out[60]: array([[ True,  True, False, False]], dtype=bool)

In [61]: b
Out[61]: array([[ True, False,  True, False]], dtype=bool)

In [62]: np.where(a*b,np.nan,(2*b + a))
Out[62]: array([[ nan,   1.,   2.,   0.]])