On the Bénabou-Roubaud theorem
Bruno Kahn
TL;DR
The paper provides a detailed proof of the Bénabou–Roubaud theorem in its original setting and shows that its hypotheses can be weakened: the base category need not have fibre products and the Beck–Chevalley condition can be weakened to a requirement that base-change morphisms be epi. It develops adjoint-chase techniques to analyze base-change morphisms $\chi$ and relates exchange to the Chevalley condition (C) and its dual, establishing equivalences and essential-surjectivity results for the Eilenberg–Moore comparison functor $K^a$ under strengthened unit-descent hypotheses. By introducing pre-descent data and the unit condition via a diagonal $\Delta$, it proves an equivalence between the algebraic and descent-data perspectives under a technical hypothesis $h_1$, and it provides a supplement showing associativity of unital $T^a$-algebras under fully faithful pullback functors accompanied by Karoubi completeness. The appendix gives a concrete case where the exchange condition holds (notably reproducing Mackey-type formulas in the group context) and discusses how these results extend to presheaf categories and representations in categorical settings, with implications for base-change and descent theory.
Abstract
We give a detailed proof of the Bénabou-Roubaud theorem. As a byproduct it yields a weakening of its hypotheses: the base category does not need fibre products and the Beck-Chevalley condition, in the form of a natural transformation, can be weakened by only requiring the latter to be epi.
