Dr' Lewis Stiller - PowerPoint PPT Presentation

1 / 57
About This Presentation
Title:

Dr' Lewis Stiller

Description:

Unmove rook up. 32 1998, 1999 by Lewis Stiller. Unmoving pieces ... Queen: Rook Bishop. Knight: Ignore holes. King: Tilt only for one time unit. 33 ... – PowerPoint PPT presentation

Number of Views:60
Avg rating:3.0/5.0
Slides: 58
Provided by: argyrisk
Category:
Tags: lewis | rook | stiller

less

Transcript and Presenter's Notes

Title: Dr' Lewis Stiller


1
Computer Endgame Analysis
  • Dr. Lewis Stiller

2
HEURISTICS VERSUS PERFECTION
  • Heuristics Tools to simplify real-life problems
  • Weather prediction
  • Game-playing
  • Stock-market analysis
  • Theorem proving
  • Language understanding
  • Perfection Only attainable in fairly small
    domains.

3
Overview
  • History
  • Method
  • Results
  • Conclusion

4
History
5
Historical Human
  • Al-Adli - 9th century
  • Carrera - 1617
  • Philidor - 1749
  • Lolli - 1763
  • Horwitz and Kling - 1851
  • Crosskill - 1864
  • Molien - 1893
  • Amelung ca. 1901
  • Troitsky - 1934

6
AL-ADLI 9TH CENTURY A.D.
7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Historical Computer
  • Quevedo - 1890, 1915
  • Zermelo - 1913
  • Bellman - 1965
  • Strohlein - 1970
  • Komissarchik and Futer - 1974
  • Herik, Herschberg - 1980s and 90s
  • Thompson - 1986
  • Michie and Bratko - 1987
  • Stiller - 1989
  • Stiller - 1991

12
(No Transcript)
13
QUEVEDOS ROBOT
Torres believes that the limit has by no means
been reached on what automatic machinery will do,
and in substantiation of his opinions presents
his automatic chess-playing machine Scientific
American Suppl. 1915
14
(No Transcript)
15
DEUTSCHE SCHACHZEITUNG Approx. 1902
16
(No Transcript)
17
REPRESENTATIONS (1897)
18
(No Transcript)
19
Read a hundred novels in a language, and you will
know that language
THE MOLIENS
20
TOMSK MANUSCRIPT
21
Method
22
Endgame computation
Set of chess pieces
Endgame algorithm
All winning positions using those pieces
23
Algorithm overview
White wins in 1
White wins in 2
White wins in 3
White wins in 4
24
Backup
Complement
White to play and win within j moves
Black unmove
White cannot win within j moves
Complement
Black can avoid loss for at least j moves
White unmove
Black loses within j moves
White wins within j1 moves
25
HYPERBOARD
(6,3)
(2,3)
(2,3,6,3)
26
HYPERBOARD
27
OPERATOR
28
HYPERBOARD OPERATOR
(4,3,6,3)
(5,3,6,3)
(3,3,6,3)
(2,3,6,3)
29
HYPERBOARD UNMOVE ROOK RIGHT
30
Unmove-rook-right
Set of positions
Marbles in hyperboard
Holes in hyperboard
Tilt hyperboard
Get points rolled over
New set of positions
31
Unmove rook up
32
Unmoving pieces
  • Rook Tilt up, right, down, left
  • Bishop Tilt diagonally
  • Queen Rook Bishop
  • Knight Ignore holes
  • King Tilt only for one time unit

33
ORBIT UNDER D4
34
D4 ACTION ON HYPERBOARD 8-ELEMENT ORBIT
35
QUOTIENT SPACE
36
Unpromotion
KQ v. K
KP v. K
37
Implementation
  • Reduced problem to manipulation of boolean arrays

38
3-Dimensional Hypercube
39
4-Dimensional Hypercube
40
5-Dimensional Hypercube
41
16-Dimensional Hypercube
42
Results
43
Stiller results
  • Five-piece games
  • Previous 2 weeks (Thompson)
  • 2 minutes
  • Six-piece games

44
Win versus Mate metric
  • WTM and mate in 5
  • WTM and win in 5
  • Win Mate or winning capture or winning pawn move
  • If WTM and win in k then WTM and mate in kgtk

45
WTM and win in 223
46
WTM and Win in 243
47
NUMBER OF MOVES REQUIRED TO WIN CERTAIN
ENDGAMES
Maximum value of the max-to-win over all
positions with White to move and win in the given
endgame.
48
6-PIECE MAX-TO-WINS
Maximum value of the max-to-win over all
positions with White to move and win in the given
endgame.
49
KARPOV-KASPAROV TILBURG,1991
50
Mutual Zugzwangs
  • White to play and draw Black to play and lose
  • Leads to elegant studies

51
Whites turn White drawsBlacks turn White wins
52
Whites turn White drawsBlacks turn White wins
53
Study by N. Elkies
54
Results
55
Conclusion
56
Conclusions
  • Chess history Theodor Molin founded two fields
    at about the same time group representation
    theory and numerical analysis of chess endgames
    (before spending the rest of his life in
    Siberia).
  • Deep and surprising results can lurk in the
    simplest settings.

57
ACKNOWLEDGMENTS
  • N. Elkies, S. Kasif, H. Berliner, D. Waltz, B.
    Wendroff.
  • Computing facilities were provided by the
    Advanced Computing Laboratory of the Los Alamos
    National Laboratory, Los Alamos NM 87545. This
    work was also supported in part by the Army
    Research Office under Contract Number
    DAALO3-C-0038 with the University of Minnesota
    Army High Performance Computing Research Center
    (AHPCRC) and the DoD Shared Resource Center at
    AHPCRC. The Author was partly supported under a
    National Defence Science and Engineering Graduate
    Fellowship, U.S. Army Grant DAAL03-929G-0345.
  • Alpha Server Computing Facilities provided by
    Digital Equipment Corporation the Author thanks
    D. Hunter and T. Mann of Digital Equipment
    Corporation.
Write a Comment
User Comments (0)
About PowerShow.com