Overview
- The Python NumPy function bitwise_and() applies bitwise & operation between the elements of the two given array_like objects whose elements are integers.
- The bitwise & operation works on the binary form of the two integers.
- Each bit of one integer is compared with the corresponding bit of the other integer. If both the bits are 1 then the resultant bit is 1. Else the resultant bit is 0. The truth table of the AND operation is given here.
- Numpy also provides the operator & which internally uses the bitwise_and() function.
- In addition to the array-like instances, the function bitwise_and() also accepts int and Boolean scalar types of NumPy as parameters along with Python int and Boolean types. If both the parameters are scalar, the function returns a scalar. NumPy scalars are single values of NumPy types and Python types, including class instances. The NumPy bitwise_and() works on array-likes of different shapes as long as the shapes are in accordance with the broadcasting rules of NumPy and returns an ndarray. Otherwise NumPy raises an Exception stating "ValueError: operands could not be broadcast together with shapes".
Truth table - AND
| Operand1 | Operand2 | Operand1 AND Operand2 |
|---|---|---|
| 0 | 0 | 0 |
| 0 | 1 | 0 |
| 1 | 0 | 0 |
| 1 | 1 | 1 |
Example
While printing, the example uses the numpy function binary_repr() to convert the elements of the ndarray into binary format. Using binary_repr() enables inclusion of padding bits before a binary number. The binary_repr() function accepts the width of the binary number which helps in printing the number with uniform number of digits. The Python built-in function bin() also converts a decimal number into binary form.
# Example Python program that applies bitwise AND operation
# between the members two ndarray instances using the function
# numpy.bitwise_and().
import numpy
# Create ndarray for numbers
numbers = numpy.ndarray(shape = (2,3), dtype=numpy.int32)
# Create ndarray for bitmasks
masks = numpy.ndarray(shape = (2,3), dtype=numpy.int32)
# For a given number return the binary number
toBin = lambda x: numpy.binary_repr(x, width=5)
# Function to fill an ndarray
def arrayInit(nparray, startVal):
count = startVal
for i in range (0, 2):
for j in range (0, 3):
count += 1
nparray[i][j] = count
# Fill numbers
arrayInit(numbers, 20)
# Fill masks
arrayInit(masks, 0)
# Print the numbers and the masks in binary form
transformBase = numpy.vectorize(toBin)
binaryNumbers = transformBase(numbers)
print("Numbers:")
print(binaryNumbers)
print("Masks:")
print(transformBase(masks))
# Apply bitwise AND
print("Result of applying bitwise AND(number & mask):")
masked = numpy.bitwise_and(numbers, masks)
# Print the results in binary form
print(transformBase(masked))
print(type(masked))Output
Numbers:
[['10101' '10110' '10111']
['11000' '11001' '11010']]
Masks:
[['00001' '00010' '00011']
['00100' '00101' '00110']]
Result of applying bitwise AND(number & mask):
[['00001' '00010' '00011']
['00000' '00001' '00010']]
<class 'numpy.ndarray'>