Documentation

LeanPool.Feige.PaperAssembly

Final assembly of the unit-slack theorem #

This module combines the Vlassis--Thomas calibration theorem (Theorem 2.1) with the simplex/exponential identification for the statistic in (2.1). The remaining parameter is the α = 0 centroid-halfspace bound used by the δ = 1 specialization of the geometric argument in §2.2.

The δ = 1 specialization of Theorem 1.1 with the probabilistic block fully discharged.