In 1953 Roth [3] proved that for any fixed δi0N>sufic,ien tly lairgef and A is any subset of {1, 2,...toNN,hefn As icozneta in>s} a nδon-trivial 3-term arithmetic progression. In the 1980s I came up with an alternate proof that is in some aspects a little simpler but which I did not publish. This school gives me another opportunity to present this approach.
展开▼