Count basic math operations

Question

I would like to get a guess on the computaional effort of my algorithms, meaning in this case: How often do I add, subtract, multiply and divide in my algorithm .

Example:

def b(a):
    a = a + 3  #1
    a = a * a  #2
    a = a * -1 #3
    return a

if __name__ == "__main__":
    for i in range(10):
        b(1)

the ouput of the analysis should be:

1: + --> 1 mathmatical operation

2: * --> 1 mathmatical operation

3: +, - --> 2 mathmatical operations

for every loop, therefore the analyzer run on this program should return 40 .

Answer 1

A Pythonic solution might be to create a class Operand, with associated operators +, -, *, /, specified with special methods (__add__, __radd__, __mult__, __rmult__, etc.). If you use the "__r" methods correctly you can start also including litterals (1, 2.0, 3.0) in pretty much any order, as long as there is one Operand.

The class would have a counter, set to zero at creation time. Each call adds 1, but also adds the counter of each operator (default = 0).

Then you could write:

a = Operand(2)
b = a+2-5
assert b.effort == 3

Then you can go ahead and calculate your effort:

a = Operand(5)
a = a + 3  #1
a = a * a  #2
a = a * -1 #3
print(a.effort)

In the first line, you create the object. In the second line, the result of a + 3 is 8, and the resulting object has an effort of 1 . In the third, it has an effort of 3 (1 + 1 for a + 1 for a). In the fourth, it has an effort of 4. Etc.

I haven't implemented and tested it, but it should work more or less according to those principles.

Note about litterals: a + 2 + 3 would give you a count of 3, but 2 + 3 + a might give you a count of 2 (since 5 is still an integer, and not an Operand). So perhaps you want to avoid the __r methods, to stay on the safe side... it depends.