dtm: (Default)
[personal profile] dtm
We visited family in northern Wisconsin over New Year's, and one of the things that happened there is that we were introduced to the show "The Big Bang Theory", and I watched the whole first season on DVD.

So we've been working our way through the second season's DVDs here the past few weeks. As you may know if you've seen the show, in one episode this season they introduce the game of "Rock, Paper, Scissors, Lizard, Spock" (a game originally described here as "Rock Paper Scissors Spock Lizard"), an extension of "Rock, Paper, Scissors". (I'm going to call this game "RPSSL" in this post)

This got me thinking more generally about extensions to "Rock, Paper, Scissors", (hereafter, RPS) so let me define:
A degree-n RPS extension, where n is an odd number >= 3, is a two-player game in which players secretly select one of n symbols and then simultaneously reveal their choices; assuming the player selected differently from each other, the winner of the game is determined by the game's ruleset. The ruleset of an RPS extension must specify that every token wins against (n-1)/2 of the other tokens, and loses against the other (n-1)/2 tokens.

So RPSSL is a degree-5 RPS extension, because for any choice player A makes, exactly two choices for player B will let A win (and exactly two will let B win).

Two RPS extensions are equivalent if you can obtain one from the other merely by renaming tokens.

So here's what I've come up with:

  • There's at least one degree-n RPS extension for every odd n >= 3.
  • Up to equivalency, there's only one degree-5 RPS extension.
  • Up to equivalency, there are at most two degree-7 RPS extensions.

I need to work more with the two different degree-7 extensions I have to prove that they definitely aren't equivalent. Getting some paper and writing it down would probably help, rather than trying to just keep it in my head.

For the other two points, I'll just note that if you re-arrange the tokens in RPSSL into the order:
Rock, Spock, Paper, Lizard, Scissors

And imagine those tokens in a circle, then every token loses to the two tokens that immediately follow it and beats the two that immediately precede it. This shows a general pattern that can work for any n and working out that with five tokens, there must be such an order proves the second point.

Date: 2011-01-23 03:34 am (UTC)
From: [identity profile] silkblade.livejournal.com
An online comic I read posted a comic about RPS recently.


Date: 2011-01-23 05:37 am (UTC)
From: [identity profile] joxn.livejournal.com
RPSes are special kinds of directed graphs, and graph theory is not my strong point, but here's a thought I had: construct a degree-pq (p,q odd) RPS as follows: make q cliques of p nodes and connect them with a degree-p RPS. Then make connections among the q cliques according to a degree-q RPS. Now do the same swapping p and q. Can we say anything interesting about the two RPSes? (I think they have to be distinct. The degree-15 RPSes might be interesting examples because 3 and 5 only have 1 possible clique-shape.)

May 2017

212223242526 27

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Sep. 19th, 2017 06:52 pm
Powered by Dreamwidth Studios