Thursday, March 14, 2013

Proof by Genie, continued

I am interested in collecting more examples of Proof by Genie, as mentioned in a previous post. If you come across an example of this, please let me know.

I've now remembered another example of proof by genie, actually from one of my own recent papers. Two people are trying to find each other on the vertices of a triangle. Initially they start at two different vertices and at each step $t=1,2,\dotsc$ each player stays or moves to another vertex, trying to minimize the expected number or steps until they meet. In working out the best symmetric rendezvous strategy it helps to assume that the players are given a common notion of what is "clockwise" on the triangle. However, once the optimal strategy is found it turns out that a common notion of clockwise is not needed to implement the optimal strategy. If you would like to learn more about this problem see 


R. R. Weber, Optimal symmetric rendezvous search on three locations, Math Oper Res., 37(1): 111-122, 2012.

or seminar slides. One of the slides begins with the words "Suppose the problem is made a bit easier: the players are provided with the same notion of clockwise. (We will see later that this does not actually help the players.)"