A bicategorical version of Masuokas theorem. Applications to bimodules over functor categories and to firm bimodules

We give a bicategorical version of the main result of A. Masuoka ({Corings and invertible bimodules,} {\em Tsukuba J. Math.} \textbf{13} (1989), 353–362) which proposes a non-commutative version of the fact that for a faithfully flat extension of commutative rings $R \subseteq S$, the relative Picard group $Pic(S/R)$ is isomorphic to the Amitsur 1–cohomology group $H^1(S/R,U)$ with coefficients in the units functor $U$.

💡 Research Summary

The paper presents a bicategorical generalisation of Masuoka’s classical theorem, which originally identified the relative Picard group Pic(S/R) of a faithfully flat extension of commutative rings R⊂S with the Amitsur first‑cohomology group H¹(S/R,U) for the units functor U. The authors replace the commutative‑flat hypothesis by a purely categorical condition: the extension must be comonadic in an appropriate bicategory. Working inside a bicategory 𝔅 that possesses the necessary (co)limits, they consider two monoid objects A and B and a 1‑cell M that plays the role of an A‑B bimodule. By equipping M with a comonad‑induced coalgebra structure they obtain a codiscent object. The main theorem asserts that the group of invertible 1‑cells (i.e., invertible bimodules) between A and B is naturally isomorphic to the relative Picard group Pic(A,B) and, simultaneously, to the Amitsur cohomology group H¹(A/B,U) defined via the standard Amitsur complex built from the comonadic adjunction.

The proof proceeds in two stages. First, the authors verify that the comonadic functor preserves the relevant limits and colimits, guaranteeing that the codiscent object is effective—the bicategorical analogue of effective descent. Second, they show that effective codiscent data correspond precisely to 1‑cocycles in the Amitsur complex; the invertibility condition on a 1‑cell translates into the existence of a 2‑cell inverse, which is exactly the cocycle condition. Consequently, the set of equivalence classes of invertible bimodules coincides with the first cohomology set.

Two substantial applications are developed.

1. Bimodules over functor categories. The authors treat the bicategory Fun(C,D) whose objects are functors C→D, 1‑cells are natural transformations, and 2‑cells are modifications. Monoid objects correspond to monoidal functors, and bimodules become functors equipped with compatible left and right actions. The comonadic condition is satisfied whenever the underlying categories admit the required (co)limits, which is the case for most familiar settings (e.g., modules, sheaves, chain complexes). The bicategorical Masuoka theorem then yields an explicit identification of the relative Picard group of such functor‑bimodules with the Amitsur cohomology computed in the functor category. This bridges the gap between classical descent theory for rings and modern descent for diagrammatic objects.

2. Firm bimodules. A firm module over a possibly non‑unital ring R is a module M such that the canonical map M⊗_R R → M is an isomorphism. The authors show that firm bimodules also fit into the bicategorical framework: the lack of a unit does not obstruct the construction of a comonadic adjunction, because the firm condition guarantees the necessary exactness properties. Consequently, the group of invertible firm bimodules between two firm algebras is again isomorphic to the corresponding Amitsur cohomology group, providing a non‑unital analogue of Masuoka’s result.

The paper includes several concrete examples: (i) the classical commutative flat case, recovered by taking the bicategory of rings and bimodules; (ii) corings and their comodules, illustrating the original motivation of Masuoka; (iii) tensor‑product functors on categories of modules, showing how the theorem operates in a functorial environment.

In the concluding section the authors emphasise the conceptual gain of the bicategorical viewpoint: the comonadic hypothesis subsumes flatness, faithfully flat descent, and many other technical conditions, allowing the theorem to apply uniformly across a wide spectrum of algebraic contexts, including non‑commutative, non‑unital, and higher‑categorical settings. They suggest further extensions to ∞‑bicategories and to non‑abelian cohomology, indicating a promising direction for future research.