Fixed point vs Floating point arithmatic

How do we instruct the compiler to use Fixed point arithmatic or floating point arithmatic in Python.
This is in relation to the accelerators for AI course, by IITR.