[Freeciv-Dev] Re: (PR#12834) amortize()

["Brian Dunstan" <bdunstan149@xxxxxxxxx>]

<URL: http://bugs.freeciv.org/Ticket/Display.html?id=12834 >


--- Jason Short <jdorje@xxxxxxxxxxxxxxxxxxxxx> wrote:
> 
> <URL:
> http://bugs.freeciv.org/Ticket/Display.html?id=12834
> >
> 
> Benoit Hudson wrote:
> > <URL:
> http://bugs.freeciv.org/Ticket/Display.html?id=12834
> >
> > 
> > On Mon, Apr 18, 2005 at 02:47:03PM -0700, Brian
> Dunstan wrote:
> > 
> >><URL:
> http://bugs.freeciv.org/Ticket/Display.html?id=12834
> >
> >>
> >>I noticed that the amortize() function uses a very
> >>clever, but complicated, algorithm.
> > 
> > 
> > This code is astounding.  Why don't we do this
> calculation in
> > floating-point?  It would be (with casts to
> doubles made beforehand):
> >         double discount = 1.0 - 1.0 /
> ((double)MORT);
> >         return benefit * pow(discount, delay);
> > or
> >         return (int) (benefit *
> pow(discount,delay) + 0.5);
> > if you want to round.
> 
> Ohmigod.
> 
> The crazy thing is, I'm sure this is done to be more
> efficient but the
> end result is it makes the calculation O(delay)
> instead of O(log(delay)).
> 
> clever, but complicated, algorithm.

This code is astounding.  Why don't we do this
calculation in
floating-point?  It would be (with casts to doubles
made beforehand):
        double discount = 1.0 - 1.0 / ((double)MORT);
        return benefit * pow(discount, delay);
or
        return (int) (benefit * pow(discount,delay) +
0.5);
if you want to round.
> -jason


Just as an aside, it looks like a lot of the original
code went to great lengths to avoid floating point
calculations.  I am not sure if avoiding floating
point  math is a goal any more, or not.






                
__________________________________ 
Do you Yahoo!? 
Plan great trips with Yahoo! Travel: Now over 17,000 guides!
http://travel.yahoo.com/p-travelguide