The bigger Brauer group is really big. (Q1810558)

There is an old question if, for a scheme \(X\), the natural map from the Brauer group, consisting of equivalence classes of Azumaya algebras, to the cohomological Brauer group, the torsion group of \(H_{\text{ét}}(X,\mathbb{G}_m)\), is surjective. The notion of Azumaya algebras has been generalized to the notion of the central separable algebras (not necessarily localy free or renital) and has been proven that on arbitrary Noetherian schemes, each Čech, 2-cohomology class comes from a coherence central separable algebra but not every 2-cohomology class comes from Čech cocycles. The author of the paper under review proves a general convergence result for étale cohomology, namely that the canonical map \[ \vec H_{\text{ét}}(X,F)\to H_{\text{ét}}(X, F) \] is bijective for any abelian sheaf \(F\), provided that each \(n\)-tuple of points \(x_1,x_2,\dots, x_n\in X\) admits an affine open neighborhood, generalizing a result of M. Artin. For Noetherian schemes such that each pair of points admits an affine open neighborhood the above result implies first that \(\widetilde {\text{Br}} (X)= H^2_{\text{ét}}(X, \mathbb{G}_m)\), where \(\widetilde {\text{Br}}(X)\) denotes the bigger Brauer group, defined as the group of equivalence classes of central separable algebras and second that the inclusion \[ H^2_{\text{Zar}}(X,\mathbb{G}_m)\subset \widetilde {\text{Br}} (X) \] holds for any normal Noetherian scheme showing at the end that there are central separable algebras neither equivalent to Azumaya algebras nor given by Čech cocycles.