TheoremComplete

Gerbes and Brauer Groups - Main Theorem

The fundamental theorems about gerbes and Brauer groups establish their classification and connect them to cohomology theory.

TheoremGerbe Classification Theorem

Let A\mathcal{A} be an abelian sheaf of groups on a site C\mathcal{C}. There is a bijection: {equivalence classes of A-gerbes on X}H2(X,A)\{\text{equivalence classes of $\mathcal{A}$-gerbes on } X\} \leftrightarrow H^2(X, \mathcal{A})

Moreover, automorphisms of a gerbe G\mathcal{G} are classified by H1(X,A)H^1(X, \mathcal{A}), and the neutral gerbe BA\mathcal{B}\mathcal{A} corresponds to 0H2(X,A)0 \in H^2(X, \mathcal{A}).

This establishes gerbes as geometric realizations of second cohomology classes.

TheoremCohomological Brauer Group

For a scheme XX, there is an exact sequence: 0Pic(X)HZar2(X,Gm)Br(X)H1(X,Pic)0 \to \text{Pic}(X) \to H^2_{\text{Zar}}(X, \mathbb{G}_m) \to \text{Br}(X) \to H^1(X, \text{Pic})

When XX is regular or quasi-projective, this simplifies to: Br(X)=Heˊt2(X,Gm)tors\text{Br}(X) = H^2_{\text{\'et}}(X, \mathbb{G}_m)_{\text{tors}}

The Brauer group is always a torsion abelian group, and for proper varieties over algebraically closed fields, it's finite.

TheoremGabber's Theorem

For any quasi-compact scheme XX with affine diagonal, the natural map: Br(X)Br(X)\text{Br}'(X) \to \text{Br}(X) is an isomorphism, where Br(X)\text{Br}'(X) is the cohomological Brauer group. This resolves a long-standing question about different definitions of the Brauer group.

TheoremGrothendieck's Purity Theorem

Let XX be a regular scheme and ZXZ \subset X a regular closed subscheme of codimension dd. Then: HZi(X,Gm)={0i<dZi=dH^i_Z(X, \mathbb{G}_m) = \begin{cases} 0 & i < d \\ \mathbb{Z} & i = d \end{cases}

For d=2d = 2, this implies that elements of Br(X)\text{Br}(X) supported on smooth divisors are zero, providing strong constraints on Brauer classes.

TheoremArtin-Mumford Theorem

There exist unirational threefolds that are not rational, with the obstruction lying in Br(X)\text{Br}(X). Specifically, they constructed a conic bundle XP2X \to \mathbb{P}^2 with Br(X)\text{Br}(X) non-trivial, showing: X(C(t))=X(\mathbb{C}(t)) = \emptyset for the function field, proving non-rationality. This was the first non-rationality result using cohomological methods.

TheoremTsen-Lang Theorem

For a field kk of transcendence degree dd over an algebraically closed field: Br(k)=0 if d1\text{Br}(k) = 0 \text{ if } d \leq 1

For function fields of curves over C\mathbb{C}, the Brauer group vanishes. This has profound consequences for rationality questions and descent theory over function fields.

TheoremAzumaya Algebra Theorem

For a scheme XX, there is an equivalence: {Azumaya algebras on X}/MoritaBr(X)\{\text{Azumaya algebras on } X\}/\text{Morita} \simeq \text{Br}(X)

Given αBr(X)\alpha \in \text{Br}(X), there exists an Azumaya algebra A\mathcal{A} with [A]=α[\mathcal{A}] = \alpha. The Morita equivalence class determines and is determined by the Brauer class, connecting non-commutative algebra to cohomology.

TheoremColliot-Thélène-Sansuc Formula

For a smooth geometrically rational surface SS over a number field kk: Br(S)/Br(k)(Q/Z)ρ\text{Br}(S)/\text{Br}(k) \simeq (\mathbb{Q}/\mathbb{Z})^{\rho} where ρ\rho is the Picard rank. This explicit formula allows computation of Brauer groups for rational surfaces and has applications to the Brauer-Manin obstruction.

These theorems establish the Brauer group and gerbes as fundamental invariants in geometry and arithmetic, with deep connections to cohomology, rationality, and descent.