Author: KarinsDad
Date: 16:19:54 07/20/99
Go up one level in this thread
On July 20, 1999 at 16:21:54, Dann Corbit wrote:
>On July 20, 1999 at 15:31:21, KarinsDad wrote:
>[snip]
>>11) Having a carefully prepared list will stifle innovation. People who WANT to
>>tinker will be told "Well, the list says that the algorithm XYZZY sucks in
>>practice, so don't bother to use it.". Essential is hardly the word I would use
>>to describe such a list.
>1. Bubble sort O(n^2) {large C constant term}
>2. Insertion sort O(n^2) {small C}
>3. Heap sort O(nlogn) {large C}
>4. Singleton's sort O(nlogn) {small C}
>5. Radix sort O(n) {but a function of the key length}
>
>Does this discourage you when you want to try a sorting routine? I have a list
>like this that is much larger. And yet I have invented a sort that is now used
>in current database systems. The list did not discourage me. The list showed
>me what current routines behave like. I can look at the routines and understand
>their strenghts and weaknesses, and also not waste time studying bubble sort.
>
Dann,
You are talking apples and oranges, but people who look at your results will be
thinking just apples.
In the example you give above, each sort has a distinctive algorithm and
requires no external elements to make it work better or worse.
In a chess program, you cannot just "pull out" the Alpha-Beta search and plug in
a PVS/NegaScout and expect your results to be consistent regardless of
evaluation routines, move generators, and other aspects of the program. The same
applies to the other elements of a chess program such as the move generator.
For example, in my program, I use a square attack table bitboard to determine
things such as square control. I also use it to determine illegal moves in the
move generator such as checks and pins. If I replace my move generator with a
different one, the new one will not be able to leverage this bitboard and I will
lose speed because of it. How can I easily compare the two? The answer is that I
cannot easily do it and that it will take some major work to figure it out.
Multiply this by the number of programs out there and the number of different
changes required and the number of different algorithms you wish to examine and
you will start to get an idea of the magnitude of the problem.
Yes, some people could use some generic list as you are discussing to determine
which approach to use first (like your sort list above), but the difficulty in
making the list both useable and accurate is MUCH larger than you seem to
casually think.
I believe this topic is the first one that Bruce and I have ever agreed on. If
that does not tell you something, I doubt anything else I can say will convince
you.
KarinsDad :)
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.