prob010: social golfer problem

proposed by Warwick Harvey
wharvey@cs.monash.edu.au

References

This benchmark is derived from a question posted to sci.op-research by bigwind777@aol.com (Bigwind777) in May 1998. It is a generalisation of the problem of constructing a round-robin tournament schedule, where the number of players in a "game" is more than two. The optimal solution for 32 golfers is not yet known.

Clearly, since a golfer plays with 3 new people each week, the schedule cannot exceed 10 weeks. The best solution of which I am aware is from Stefano Novello ( S.Novello@ic.ac.uk) who posted a 9-week solution, along with the source of the ECLiPSe program used to find it. The ``challenge'' therefore is to find a 10-week solution, or prove none exists.