speeding up Python script

[Dan Sommers]

On Wed, 18 May 2005 12:56:50 +0100,
"Luis P. Mendes" <luis_lupe2XXX at netvisaoXXX.pt> wrote:

> The reason why I'm using six nested for loops is because I need to find
> the best output using those six variables as input.

> Here's the simplified code:

> for per in range():
> ~    for s in range():
> ~        for t in range():
> 	    for v in range():
> ~                for n in range():
> ~                    for l in range():
> 			var a, \
> 			var b, \
> 			...
> 			var 15 = function1(arg1, \
> 						...
> 						arg20)
> ~ 						var a, \
> 			var d, \
> 			var a, \
> 			...
> 			var 14 = function2(arg1, \
> 						...
> 						arg25)
> 			var c, \
> 			...
> 			var 18 = function3(arg1, \
> 						...
> 						arg20)
> 		ia = var1*var2-numPerdidos*sl
> 												result.insert(index, [per,s,t,v,n,l,ia])
> 		index = index + 1

> thanks for the replies

I don't know how large your search-space is, but if the ranges for s, t,
v, n, and/or l don't change, at the very least you can pre-compute those
ranges instead of creating a new list with range every time:

srange = range( <whatever> )
trange = range( <whatever> )
vrange = range( <whatever> )
nrange = range( <whatever> )
lrange = range( <whatever> )

for per in range( <whatever> ):
  for s in srange:
    for t in trange:
      for v = vrange:
        for n in nrange:
          for l in nrange:
            somevar = somefunction( per, s, t, v, n, l )
            result.append( [ somevar, per, s, t, v, n, l ] )

(Appending the results is probably quicker than inserting them, but I
don't know for sure.)

But a better-than-an-exhaustive-search algorithm sounds like a good
idea, too.

Regards,
Dan

-- 
Dan Sommers
<http://www.tombstonezero.net/dan/>