[seqfan] Re: What do you call an algorithm that is inefficient overall in some cases?
Alonso Del Arte
alonso.delarte at gmail.com
Tue Feb 9 01:50:58 CET 2010
Heuristics to me suggests some degree of sophistication. Myopic makes sense
to me because at each step the algorithm only looks at the most efficient
way to proceed at that step without concern that it could take more steps
than are necessary. Now I'm not sure what exactly it was in the dictionary
definition that made me doubt that it was the right word to use.
On Mon, Feb 8, 2010 at 5:19 AM, Olivier Gerard <olivier.gerard at gmail.com>wrote:
> Heuristics has merits and I think myopic is close to what you describe but
> you could take one of these other point of views:
>
> - This is not an algorithm for what you are searching for but for another
> not completely specified sub problem that it solves satisfactorily. Your
> new quest is now to characterize this particular problem.
>
> - This is an incorrect algorithm for the problem you want. It is not just
> inefficient (hinting that it solves but using too many resources) but has
> incorrect outputs.
>
>
> On Mon, Feb 8, 2010 at 07:03, <franktaw at netscape.net> wrote:
>
> > I'm not sure what "inefficient overall in some cases"; but "heuristic"
> > may capture the essence of what you are asking.
> >
> > Franklin T. Adams-Watters
> >
> > -----Original Message-----
> > From: Alonso Del Arte <alonso.delarte at gmail.com>
> >
> > What do you call an algorithm that is inefficient overall in some
> > cases, but
> > never at the level of an individual step? For example, the algorithm
> > described in A112687 fails to find 2-square solutions for numbers like
> > 32
> > and 244. But it does work for numbers like 50 and 82. I've looked up
> > both
> > "greedy algorithm" and "myopic algorithm" in the Oxford Math Dictionary
> > but
> > neither term seems to apply here.
> >
> > Al
> >
> >
> > _______________________________________________
> >
> > Seqfan Mailing list - http://list.seqfan.eu/
> >
>
> _______________________________________________
>
> Seqfan Mailing list - http://list.seqfan.eu/
>
More information about the SeqFan
mailing list