Why float objects in Python doesn't have denominator attribute, while int does?

Question

While I was messing around with Python,

>>> [attr for attr in dir(1) if not attr.startswith('_')]
['bit_length', 'conjugate', 'denominator', 'imag', 'numerator', 'real']
>>> [attr for attr in dir(1.1) if not attr.startswith('_')]
['as_integer_ratio', 'conjugate', 'fromhex', 'hex', 'imag', 'is_integer', 'real']

Although I understand that 'conjugate', 'imag', and 'real' are there for the sake of compatibility with complex type, I can't understand why 'numerator' and 'denominator' exists for int only, and doesn't for a float.

Any explanation for that ?

Answer 1

This is most likely because floats are somewhat lossy - they can not perfectly represent every value. Consider this example:

>>> 1.0/5.0
0.20000000000000001

If you wanted the access the denominator of 1.0/5.0 python would have to return 18014398509481984 (20000000000000001/100000000000000000 == 3602879701896397/18014398509481984). The loss of precision will cause python to have no choice but to return crazy values, so the designers chose not to implement the function.

Answer 2

This is because int is a subclass of rational, and float is not. Since rational has a denominator attribute, int inherited it.

You can read more here

Answer 3

Take a look at the number class hierarchy: Python numbers

numbers.Integral is a sub class of numbers.Rational

It's numbers.Rational that adds the numerator and denominator members.