Physically
Simulating the Name Game
We
end with descriptions of ways to physically simulate the Name Game. We can modify the biased coin model to use
transparent spinners and a variety of bases to get data about cycle lengths in
random permutations. For the first spin,
we use a base labeled from 1 to 
,
with equal regions for each label.
Suppose, for example, that 
and we spin a 4. Then, since each length from 1 to 6 is
equally probable for the length of the next cycle, we spin next using a base
labeled from 1 to 6 to determine its length, and so on.
We can also
use a deck of cards labeled from 1 to 
,
letting the first cycle end with card 
,
which has equal probability of being anywhere in the deck. The next cycle then ends with the greatest
card still in the deck after card 
has been removed, etc. This is a slight variation of a simulation
described in Golomb and Gaal (1992).
We can also
use cards to gather data on fixed points by counting from 1 to 
,
while dealing out the cards and recording the number of times the number spoken
matches the number on the card. This is
a simulation of the problemes des rencontres or the “problem of the
matches,” which was one of the first historical appearances of fixed points of
permutations. The main problem with
using cards to simulate is making sure that they are sufficiently shuffled to
be random (see, for example, Morris, 1998).
Golomb, S.W. and Gaal, P. “On
the Number of Permutations of 
Objects with Greatest Cycle Length 
.”
Probabilistic Methods in Discrete
Mathematics : Proceedings of the Fourth International Petrozavodsk Conference, Petrozavodsk, Russia, May 12-15, 1992. Utrecht, the Netherlands: VSP, 1997.
Morris, S. Brent. Magic
Tricks, Card Shuffling, and Dynamic Computer Memories. Washington
DC: Mathematical Association of America,
1998.
Note: This is a
supplement to, Borkovitz, Debra, “The Name Game: Exploring Random
Permutations.” Mathematics Teacher 98
(October 2005).
PDF Version Back to Articles Home