Author: Dann Corbit
Date: 16:42:23 05/15/01
Go up one level in this thread
On May 15, 2001 at 19:29:24, Martin Schubert wrote:
[snip]
>Some algorithms sort in O(n*n), some in O(n*log n). You say the difference is
>unimportant?
I don't know of any practical algorithm that sorts in O(n*n) {Though I know some
joke alrgorithms that are even worse}.
I would love to get a copy of such a thing to add to my collection.
Be that as it may, the point at which a demonstration was attempted was this:
If we say that an algorithm with limited input is O(1) because it will complete
in some finite time, then all algorithms are O(1). This is not a particularly
useful definition from the standpoint of practical computation, but it is
interesting mathematically (I suppose). The previous poster was making the
opposite point -- that it is useful to characterize algorithms as to how they
perform for a given change in n. And that conversely, it is less useful to
simply say that since the input is finite and the algorithm terminates, it is
O(1).
At this point, I *truly* regret starting this thread. It [the semantics
arguments] won't make our chess programs run any better and it doesn't seem to
want to tail out.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.