Question

There are 2 main reasons why interpreted Python code is slower than code in a compiled lanauge such as C (or other compiled langauge):

• Python executes byte code in a virtual machine (minor effect) while C compiles down to machine instructions specific for the processor
• Python has dynamic typing (major effect) while C is statically typed. In a dynamically typed language, the simple expression a + b can mean many, many different things, and the interrpeter has to figure out which interpretation is intended. In contrast, a and b must have a type in C such as double and there is no ambiguity about what + means to resolve.

If speed is critical, it is often necessary to exploit the efficiency of compiled languges - this can be done while retaining the nice features of Python in 2 directions

• From C to Python
• From Python to C

Here we will look at how to go from C (C++, Fortran, Julia) to Python,

def python_fib(n):
    a, b = 0,  1
    for i in range(n):
        a, b = a+b, a
    return a

%timeit python_fib(100)

100000 loops, best of 3: 8.47 µs per loop