Author: Robert Hyatt
Date: 08:55:28 02/06/06
Go up one level in this thread
On February 06, 2006 at 09:45:40, chandler yergin wrote:
>It still can't beat Human recognition of Long Term Positional Defects
>in a position, which a Computer may never find. Not from the Root position
>which is ALWAYS the position on the Board!
>In Game play Hash is Trash!
I see it is your intention to go down the _same_ road you have gone down many
times previously... that is, arguing about a point that you have zero
understanding of.
Do one of the two following things:
(1) set up a position with any program you choose. Crafty will certainly be ok
for this. set the search depth to something that takes about 2 minutes total
time. Run the position with the default hash, and write down the time required
to finish that search. Double the hash and repeat. Double and repeat.
Continue doubling until the search time no longer decreases. Note that you
should use a normal middlegame position here, not a tactical position, not an
endgame position, as those magnify the problem and the typical case is more
important. Look at the search times you got, and then decide whether more hash
is trash, or whether more hash lets the program complete the same search in less
time, making it faster. Once you do this, you won't make such statements again.
(2) simply shut up about this since you do _not_ know what you are talking
about. If you do (1) above, you will actually learn something important.
>
>The game tree consists of all possible moves for the current players starting at
>the root and all possible moves for the next player as the children of these
>nodes, and so forth, as far into the future of the game as desired. The leaves
>of the game tree represent terminal positions as one where the outcome of the
>game is clear (a win, a loss, a draw, a payoff). Each terminal position has a
>score. High scores are good. For example, we may associate 1 with a win, 0 with
>a draw and -1 with a loss.
This is not the game tree chess programs actually traverse however. This is a
pure tree without benefit of alpha/beta pruning, search extensions and search
reductions.
>Bounded Lookahead in Large Tree
>In large trees, it is quite impossible to search all the nodes. The next best
>thing is to trim the tree to a few levels and pretend that it is a good
>approximation of the (unknown) minimax tree by assigning scores to its leaves.
>The difference now is that the scores are no longer exact, but only educated
>guesses. The scores obtained in this manner are said to be calculated with the
>aid of an evaluation function. Evaluation functions are constructed by the user
>based upon insight and experience. We may still employ the minimax algorithm to
>compute all the scores:
>
>
>http://www.seanet.com/~brucemo/topics/minmax.htm
>Quoting:
> A game of chess can be considered as a (usually) large n-ary tree ("n-ary"
>means that from each node in the tree, there are some arbitrary number of
>branches leading to other nodes). The position that is on the board now is the
>root position or root node. Positions that can be reached in one move from the
>root position are reached by branches from the root position. These positions
>are called successor positions or successor nodes, Each of these successor
>positions has a series of branches emanating from it, each of which represents a
>legal move from that position.
>http://www.rebel.nl/ches2010.htm
>Recent Computer Chess developments
>
>Especially the last years ideas based on the BERLINER experiment have been
>practiced in chess programs with extremely good results in the computer-computer
>area. Chess programmers discovered that by a new approach that looks illogical
>at first sight their programs became a lot stronger in the computer-computer
>area.
>
>Computer chess development of the last years:
>Make the program as fast as possible.
>Even remove existing chess knowledge to ensure a fast chess program.
>Add lots of tactics to out-search the opponent.
>Add only the very basic chess knowledge (Berliner concept)
>Make the program aggressive to ensure tactics in games.
>Schröder about this new trend:
>
>It's my opinion it all seem to work in the comp-comp area.
>
>Now for comp-comp lovers this is great news but what about the people who use a
>chess program for analysis, study and playing games?
>
>10 days ago I released Rebel10.0c as an engine update for Rebel10.
>
>In 10c I removed some chess knowledge which made Rebel 30% faster, next this 30%
>speed gain was used to add new tactics. As a result 10c is a better comp-comp
>player now but a lower positional player than the original Rebel10.
>
>This is actually what is happening the last years and I want to report it in all
>its details in the hope it will be understood.
>
>There are advantages and disadvantages. The advantages are for the comp-comp
>lovers. The disadvantages are for people who use a chess program for analysis,
>study and playing games as the positional understanding of chess programs will
>drop and no progress is made.
>
>My personal opinion is that the disadvantages are bigger than the advantages
>because the vast majority of people that use a chess program use it for playing
>games, analysis and study.
>
>It's my wish the computer chess press will pick up the subject, explain and
>judge.
>
>Here is what your Engine is doing; how it thinks...
>A Real Game in Real Time.
>Who wants to dispute this?
>It Proves the above!
>You can see that with every move made on the Board, becomes the Root position
>and the search & analysis begins again from Ply 1.
And this has exactly _what_ to do with hash sizing???
<massive snip of irrelevant chess program output>
This page took 0.01 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.