It turns out that Behrend's construction also provides examples of sets of size , which does not contain any non-trivial solution of the equation

This is because if k points are on a sphere then their arithmetic mean cannot be on the same sphere unless they are all equal. Are there essentially better upper bounds for the size of such sets A for large k, then for k=2 that is for 3-term arithmetic progressions?

< Prev | Next > |
---|