Paradoxes in Logic, Mathematics and Computer Science - PowerPoint PPT Presentation

1 / 17
About This Presentation
Title:

Paradoxes in Logic, Mathematics and Computer Science

Description:

Two Players: This contrasts with one-player games, e.g. Sudoku. ... No Chance: The rules of the games are deterministic, unlike card games. ... – PowerPoint PPT presentation

Number of Views:236
Avg rating:5.0/5.0
Slides: 18
Provided by: wafiklo
Category:

less

Transcript and Presenter's Notes

Title: Paradoxes in Logic, Mathematics and Computer Science


1
(No Transcript)
2
Lecture 2
  • More Finite Games

3
All games considered in Lectures 1 and 2 have the
following features
  • Two Players This contrasts with one-player
    games, e.g. Sudoku.
  • Win-Lose A player wins iff the other loses.
    (except for chess, which can be modified to be a
    win-lose game). Other games could be win-win.
  • Perfect Information The positions and rules of
    the game are known to both players.
  • No Chance The rules of the games are
    deterministic, unlike card games.
  • Finite Game The game does not last forever.
  • Finite Options The players options are finite.

4
An Illustrative Example
  • Board The 0-1 list
  • Two Players Left (L) and Right (R). L moves
    first, then R, then L, then R.
  • Rules 1) Each player splits the list into 2
    equal halves and removes one of them. 2)
    When only one cell is left, L wins iff the
    cell contains 1.
  • Question How should L play?

5
The Max-Min Tree
  • This is a tree of Game Positions.
  • The root is the initial position.
  • The children of a node indicate the possible
    options of the player who plays next.
  • The leaves are the terminal (end) positions.
  • Leaves are labeled with the gain for L.
  • The gains of the children carry over to the gain
    of the parent by the following rule For each
    node n,
  • If L plays, then gain(n) max(gains(children(n)))
    .
  • If R plays, then gain(n) min(gains(children(n)))
    .

6
Exercises
  • For the following games
  • 1) Construct a max-min game tree and
  • 2) Decide if L has a winning strategy.
  • The Chocolate Game with 7 squares.
  • Nim (3 heaps) from the position (1,2,3)
  • 3?3 Tic-Tac-Toe (X-O). Here L starts by placing X
    on a 3?3 grid, R places O, etc.. The winner is
    the first one occupying a row, a column or a
    diagonal. Note This is quite a large tree,
    which could be reduced using symmetry. Also, you
    can draw different subtrees on separate pages.

7
Theorem In a finite win-lose game, exactly one
of the players has a winning strategy.
  • We claim that one of the players can always win,
    no matter how the other plays.
  • Proof
  • Its impossible that both players have winning
    strategies.
  • If L (say) does not have a winning strategy, a
    winning strategy for R is to just play in a way
    that prevents L from a winning strategy.
  • Since the game eventually ends, R wins.

8
Theorem In a finite win-lose game, exactly one
of the players has a winning strategy.
  • Another (rigorous) Proof Since the game is
    finite, its Max-Min tree has a finite height.
  • By induction on the height of that tree, we can
    show that at any node n
  • 1) L has a winning strategy iff gain(n) 1.
  • 2) R has a winning strategy iff gain(n) 0.
  • Its also obvious (by induction again) that
    gain(n) 0 or 1. Thus, exactly one of L or R has
    a winning strategy.

9
The Hex Game
  • Board An 11?11 rhombus of hexagons.
  • Two opposing sides are colored red and the others
    blue.
  • Two Players Red (R) and Blue (L). Red moves
    first, etc..
  • Rules
  • 1) Each player places a red/blue stone on a cell.
  • 2) The first player who forms a connected path
    linking his/her opposing sides wins.

10
Initial Position
11
Final Position Red wins
12
Theorem Red can always win.
  • Proof
  • The game never ends in a tie (draw), i.e. Red
    wins iff Blue loses.
  • Exactly one of them has a winning strategy.
  • If Blue has a winning strategy, then Red can
    steal it by first making an arbitrary move, then
    following Blues strategy. When Red needs to
    place a stone on an occupied place, she makes
    another arbitrary move, etc..

13
What about Chess?
  • Position An 8?8 chess board with a configuration
    of the pieces.
  • Two Players White and Black. White moves first,
    etc..
  • Initial Position

14
Can the players play forever?
  • Answer No.
  • One of the rules states that the game ends in a
    draw, if the same board position recurred 3
    times. Since the number of positions is finite,
    some positions will recur infinitely often.
  • Also, another rule states that a draw occurs, if
    at least fifty moves (by each side) have passed
    with no pawn being moved and no capture being
    made.

15
Two Open Questions
  • Question Does White have a winning strategy?
  • Answer Unknown.
  • Question Does Black have a strategy that results
    into a win or tie?
  • Answer Unknown.
  • However Exactly one of the above questions has
    the answer yes.

16
How should the players play?
  • Chess is too complicated
  • Number of legal chess positions ? 1043
  • Size of the game tree ? 10120
  • Contrast with the number of atoms in the universe
    (between 4?1079 and 1081)
  • No simple mathematical (known) theory.
  • What we can do is to search the tree of moves up
    to depth n, where for professional chess n ? 8.

17
  • Thank you for listening.
  • Wafik
Write a Comment
User Comments (0)
About PowerShow.com