MathGroup Archive 2005

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Timing of looping operators

  • To: mathgroup at smc.vnet.net
  • Subject: [mg62420] Re: [mg62416] Timing of looping operators
  • From: "Hermann Schmitt" <schmitther at t-online.de>
  • Date: Wed, 23 Nov 2005 06:27:31 -0500 (EST)
  • References: <200511230931.EAA20248@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Hello Daniel,
loops are normally more efficient then recursive function calls, because in
function calls more temporary variables have to be created.
In my oo system I first tried recursive function calls, but then changed to
loops.
Hermann Schmitt
----- Original Message -----
From: "dh" <dh at metrohm.ch>
To: mathgroup at smc.vnet.net
Subject: [mg62420] [mg62416] Timing of looping operators


> Hello,
> The Mathematica gospel tells you that you should NOT use loops because
> it is inefficient.
> Well consider the following examples and their times:
>
> n=10^6;
> d=Table[i,{i,n}];
> fun[x_]:=x;
>
> a)  Timing[fun & /@ d;]       needs 0.8 sec
> b)  Timing[Scan[fun, d]]      needs 1 second
> c)  Timing[Do[f[i], {i, n}];] needs 0.7 sec
>
> a) applies the function and creates a new list. b) does not create a new
> list -- but it is slower! And finally c) the loop is fastest!!!
>
> If you change the function to: f[x_]:=x^2, the times are even more
striking:
>
> 0.8, 2.4, 0.7
> it seems like in a and c the function evaluation takes negliable time
> compared to the loop construct, but not so in b.
>
> has anybody an explanation???
>
> Daniel
>


  • Prev by Date: permutations
  • Next by Date: Re: Solving an integral in the limit.
  • Previous by thread: Timing of looping operators
  • Next by thread: Re: Timing of looping operators