he urakperso.ens-lyon.fr/edouard.bonnet/talk/durak-poster.pdf · 2020. 7. 30. · the complexity of...
TRANSCRIPT
THE COMPLEXITY OF PLAYING DURAKÉDOUARD BONNET
INTRODUCTIONThe complexity of board games has been
quite extensively studied since the dawn ofcomplexity theory. However, despite someworks on Whist and Bridge [5, 2], Uno [3],and algorithmic questions related to Set [4]and Hanabi [1], there is few known about thecomputational complexity of card games.It is still a vastly unexplored line of re-search with connections to combinatorialgame theory and parameterized complex-ity.
We tackle the russian game Durak whosegameplay is significantly different from thecard games mentioned above.
RULES OF DURAKSimplified rules with 2 players, no trump
suit, and an empty drawing pile:
• goal: getting rid of all one’s cards.
• round: sequence of moves between theattacker and her opponent the defender.
• attacker’s moves: play a card matchingthe rank of a played card.
• defender’s moves: defend with a highercard in the same suit.
• if the defender ceases to defend, hetakes all the cards in his hand; he re-mains the defender for the next round.
• otherwise, all the cards are discarded;defender/attacker switch roles.
ON DEFENDING AN ATTACK
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
sx1 sx2 sx3 sx4 sx5 sx6sC1sC2sC3sC4sC5sC6sC7sC8
WEAKNESS AND STRONG SUIT• weakness: rank where one player has
only cards dominated by cards ofher/his opponent.
• well-covered weakness: weakness suchthat all the dominating cards are ofrank not owned by the player with theweakness.
• strong suit: only owned by one player.
ON FINDING OPTIMAL PLAY
∀x0
∃x1
∀x2
∃x3
∀x4
. . .
. . .
. . .
. . .
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
s10 s20 s30 s12 s22 s32 s14 s24 s34s11 s21 s31 s41 s13 s23 s33 s43sC′
1sC′
2sC′
3sC′
4sC′′
1sC′′
2sC′′
3sC′′
4s1w . . .s2mw sO sP s1,1d. . . . . . s2d s3d s4d
OPEN QUESTIONS
• Is single-suit Durak tractable?
• What about Durak with a boundednumber of suits?
• In Durak with more than two players,games are not polynomially bounded.Is deciding if one player has a winningstrategy still in PSPACE?
• Is deciding if one player can defend anattack in NP?
• What is the complexity of Durak witha bounded number of ranks?
• Removing the threshold feature in thePSPACE-hardness construction.
REFERENCES
[1] J. Baffier, M. Chiu, Y. Diez, M. Korman, V. Mitsou, A. vanRenssen, M. Roeloffzen, and Y. Uno. Hanabi is np-complete,even for cheaters who look at their cards. In 8th InternationalConference on Fun with Algorithms, FUN 2016, June 8-10, 2016,La Maddalena, Italy, pages 4:1–4:17, 2016.
[2] E. Bonnet, F. Jamain, and A. Saffidine. On the complexity oftrick-taking card games. In IJCAI 2013, Proceedings of the 23rdInternational Joint Conference on Artificial Intelligence, Beijing,China, August 3-9, 2013, 2013.
[3] E. D. Demaine, M. L. Demaine, N. J. A. Harvey, R. Uehara,T. Uno, and Y. Uno. UNO is hard, even for a single player.Theor. Comput. Sci., 521:51–61, 2014.
[4] M. Lampis and V. Mitsou. The computational complexityof the game of set and its theoretical applications. In LATIN2014: Theoretical Informatics - 11th Latin American Symposium,Montevideo, Uruguay, March 31 - April 4, 2014. Proceedings,pages 24–34, 2014.
[5] J. Wästlund. A solution of two-person single-suit whist. TheElectronic Journal of Combinatorics, 12(1):R43, 2005.
RESULTS• Deciding if one player can defend until
the end is NP-hard.
• Deciding if one player has a winningstrategy is PSPACE-complete.
Question for you: why is the game lengthpolynomially bounded?