Difference between revisions of "Type B Strategy"
GerdIsenberg (talk | contribs) (Created page with "'''Home * Search * Type B Strategy''' The '''Type B Strategy''', proposed in 1949 by Claude Shannon in his groundbreaking publication ''Programming a Co...") |
GerdIsenberg (talk | contribs) |
||
Line 45: | Line 45: | ||
* [[Selectivity]] | * [[Selectivity]] | ||
* [[Type A Strategy|Shannon's Type A Strategy]] | * [[Type A Strategy|Shannon's Type A Strategy]] | ||
+ | |||
+ | =Forum Posts= | ||
+ | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=68823 B strategy is the future] by [[Thorsten Czub]], [[CCC]], November 04, 2018 | ||
=External Links= | =External Links= |
Revision as of 17:11, 12 December 2018
Home * Search * Type B Strategy
The Type B Strategy, proposed in 1949 by Claude Shannon in his groundbreaking publication Programming a Computer for Playing Chess [1], is a selective approach to search minimax trees considering only a subset of plausible moves in contrast to the brute-force Type A strategy.
Quotes
from Shannon's Programming a Computer for Playing Chess:
From these remarks it appears that to improve the speed and strength of play the machine must:
-
Examine forceful variations out as far as possible and evaluate only at reasonable positions, where some quasi-stability has been established.
-
Select the variations to be explored by some process so that the machine does not waste its time in totally pointless variations.
A strategy with these two improvements will be called a type B strategy. It is not difficult to construct programs incorporating these features. For the first we define a function g(P) of a position which determines whether approximate stability exists (no pieces en prise, etc.). A crude definition might be:
| 1 if any piece is attacked by a piece of lower value, g(P) = / or by more pieces then defences of if any check exists \ on a square controlled by opponent. | 0 otherwise.
Using this function, variations could be explored until g(P)=0, always, however, going at least two moves and never more say, 10.
Type B programs
Most early chess programs were Type B and used a selective width for a maximum amount of plausible moves to be tried. Bernstein used {7, 7, 7, 7}, later programs chose width dependent from depth, Kotok-McCarthy had {4, 3, 2, 2, 1, 1, 1, 1, 0, 0}, while Greenblatt's Mac Hack used {15, 15, 9, 9, 7, ...}, and CHAOS carried out a selective search with iterative widening. With the advent of brute-force alpha-beta, and programs like Tech, Kaissa and Chess 4.5 in the early 70s, the era of the former dominating Type B programs drew to a close. Today most programs are closer to Type A, but have some characteristics of a Type B due to selectivity.
- Awit
- Black Knight
- Blitz
- CHAOS
- Chess Simulator
- Coko
- Genie
- Kotok-McCarthy-Program
- Mac Hack
- MAX (Gillogly)
- NSS
- Schach (US)
- The Bernstein Chess Program
and
See also
Forum Posts
- B strategy is the future by Thorsten Czub, CCC, November 04, 2018
External Links
- Subject: brute-force vs. intuition in math & chess by Bill Dubuque, August 1996
- Brute-force search from Wikipedia