Python: how to make if else for loops faster, when first value differs from any other?

Question

I am dealing with a function that looks like this:

def A(x):
   A=range(n)
   A[0]=(bx[0] if condition1 else cx[0])
   for i in range(1,n):
            A[i]=((dx[i] if condition2 else ex[i])

   return  map(lambda x: x+3, A)

where A is a list and b,c,d,e are operations that take x as variable. Basically, I need to make an if statement for the first value in the A-list, and a different if statement for any other value apart from the first one. Is there a way to make this more efficient?

thank you

With [`numba`](http://numba.pydata.org/) it's possible to optimise loops. Can you be more specific about b,c,d,e operations? — jezrael, Jun 12 '19 at 06:07
b,c,d,e are functions like: min(x*6/3600, 100-x/0,5), or (x*6, x-3), etc....I didn't want to over complicate the question, but they are not complex functions — Luca91, Jun 12 '19 at 06:14
no particular reason, I was trying different things to end up with a list. — Luca91, Jun 12 '19 at 06:17
I need to call the same function multiple times and within other loops and constraints, therefore I defined A(x) as a function — Luca91, Jun 12 '19 at 06:33

score 3 · Accepted Answer · answered Jun 12 '19 at 06:28

3

You can use np.where

import numpy as np
def A(x):
   A = np.arange(n)
   A[0] = (bx[0] if condition1 else cx[0])
   A[1:] = np.where(condition2, dx[1:], ex[1:])

   return A + 3

answered Jun 12 '19 at 06:28

Priyatham

2,821
1
19
33

RomanPerekhrest · Answer 2 · 2019-06-12T06:36:42.353

2

In terms of code restructuring - at least 2 optimizations can be applied in that context:

prevent double range() call
prevent redundant looping caused by map call (just to add 3 to each item)

Optimized version:

def func(x):
   res = [((bx[0] if condition1 else cx[0]) if i == 0 
          else (dx[i] if condition2 else ex[i])) + 3 
          for i in range(0, n)]

   return res

edited Jun 12 '19 at 06:36

answered Jun 12 '19 at 06:21

RomanPerekhrest

88,541
4
65
105

You may want to start with `res = [...]` instead of appending a single value to an empty list. But at least you now have working code :-) – Martijn Pieters Jun 12 '19 at 06:32
Also, a list comprehension would perform better than repeated appends. – Martijn Pieters Jun 12 '19 at 06:32
@MartijnPieters, I know it goes faster, just went for readability. Ok, let's do it – RomanPerekhrest Jun 12 '19 at 06:37

Python: how to make if else for loops faster, when first value differs from any other?

2 Answers2