Brauer Space: A Homotopical Refinement

• Brauer space is a space-level refinement of the classical Brauer group that encodes higher automorphism data and organizes Azumaya algebras into a moduli space.
• It unifies concepts from Morita theory, Picard groups, and unit groups through geometric realization, especially in commutative rings and structured ring spectra.
• Brauer spaces serve to classify étale twists and detect vanishing phenomena in moduli theory using tools like spectral sequences and extended homotopy invariants.

Searching arXiv for recent and foundational papers on Brauer spaces and related Brauer-group constructions. Brauer space is a space-level refinement of Brauer-theoretic invariants. In current literature the term is used in several closely related senses: as the geometric realization refining the classical Brauer group of a commutative ring or structured ring spectrum; as a twisted Brauer space classifying étale twists of a fixed algebra, scheme, or derived category; and as an extended Brauer space attached to a spectral algebraic stack. In algebraic geometry the same vocabulary also appears in a broader moduli-theoretic sense, where one speaks of a space parameterizing Brauer classes or interprets vanishing of the Brauer group as the absence of gerbe-theoretic obstructions to universal objects (Szymik, 2011, Antieau, 2012, Chough, 2020, Biswas et al., 2010).

1. Classical Brauer theory and the move from groups to spaces

The classical starting point is the Brauer group of a ring or scheme. For a commutative ring $R$, the Brauer group $\mathrm{Br}(R)$ consists of Morita equivalence classes of Azumaya $R$-algebras, with group law induced by tensor product. For an algebraic space or scheme $X$, the cohomological Brauer group is

$$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$

while the geometric Brauer group $\mathrm{Br}(X)$ consists of classes represented by Azumaya algebras or, equivalently, by $\mathbf{PGL}_n$-torsors for some $n$ (Szymik, 2011, Mathur, 2020).

This passage from a set or group of equivalence classes to a space is motivated by the same principle that leads from $K_0$ to algebraic $K$-theory spaces. In the Brauer setting, the point is to retain not only the connected components—corresponding to Brauer classes—but also higher automorphism data. In the discrete commutative-ring case, this recovers the Picard group and the group of units as higher homotopy groups; in spectral and derived settings, the higher homotopy is richer and reflects the higher structure of the ambient geometry (Szymik, 2011, Chough, 2020).

A recurrent theme across the subject is the comparison between geometric and cohomological Brauer theories. Several of the modern space-level constructions are designed precisely so that classes in the relevant cohomological Brauer group are represented by Azumaya algebras or twisted module categories under suitable hypotheses. This suggests that Brauer spaces are best understood as moduli spaces of Azumaya-type objects together with their higher symmetries, rather than merely as a repackaging of a torsion cohomology group (Chough, 2020, Antieau, 2012).

2. Brauer spaces and spectra for commutative rings and ring spectra

For a commutative ring $\mathrm{Br}(R)$0, Szymik defines a Brauer space by first introducing the category $\mathrm{Br}(R)$1. Its objects are Azumaya $\mathrm{Br}(R)$2-algebras; the superscript $\mathrm{Br}(R)$3 records the standard finiteness and homological conditions, and morphisms are $\mathrm{Br}(R)$4-linear equivalences of module categories, with higher simplices encoding natural isomorphisms. The Brauer space is then

$\mathrm{Br}(R)$5

the geometric realization of the nerve of this category, and the associated Brauer spectrum is

$\mathrm{Br}(R)$6

obtained from the symmetric monoidal structure by Segal’s machine (Szymik, 2011).

The basic structural theorem identifies the homotopy groups: $\mathrm{Br}(R)$7 Thus the connected components recover the classical Brauer group, the fundamental group recovers the Picard group of invertible $\mathrm{Br}(R)$8-modules, and the second homotopy group recovers the group of units. The components are naturally equivalent, as infinite loop spaces, to deloopings of the Picard space; each path component is $\mathrm{Br}(R)$9 (Szymik, 2011).

The same construction extends to structured ring spectra. In that setting, one again defines Brauer spaces and spectra using Azumaya $R$0-algebras in the brave new sense. The component of the unit is equivalent, as an infinite loop space, to the Picard space of $R$1, and the Brauer space becomes a non-connected delooping of that Picard space. Szymik describes this as a two-fold non-connected delooping of the spectrum of units. Unlike the discrete-ring case, the spectral Brauer space need not be $R$2-truncated, because the Picard space and spectrum of a commutative $R$3-algebra generally have nontrivial higher homotopy (Szymik, 2011).

This formulation is conceptually important because it packages Azumaya algebras, Morita theory, invertible modules, and units into a single homotopy-theoretic object. The Brauer group appears only as $R$4; the full Brauer space remembers the automorphism theory of the corresponding module categories.

3. The twisted Brauer space and étale-local Morita theory

Antieau’s twisted Brauer space generalizes the classical Brauer space from “forms of the unit object” to “forms of a fixed algebra or category.” Given a commutative ring or ring spectrum $R$5 and an $R$6-algebra $R$7, the $R$8-twisted Brauer sheaf $R$9 is the sheaf of spaces on the étale site of $X$0 such that, for an $X$1-algebra $X$2,

$X$3

Its connected components classify derived Morita equivalence classes of $X$4-algebras that are étale locally derived Morita equivalent to $X$5 (Antieau, 2012).

The homotopy sheaves are explicitly computed: $X$6 Here $X$7 is the sheaf of autoequivalences of the derived module category, and the higher homotopy sheaves are governed by Hochschild cohomology. This makes the twisted Brauer space a classifying space for étale-local derived Morita forms together with the higher deformation-theoretic invariants carried by Hochschild cohomology (Antieau, 2012).

A principal computational tool is the fringed spectral sequence

$X$8

For smooth and proper algebras, or ordinary rings, this collapses at a finite stage in the examples treated in the paper. The theory also yields a fiber sequence

$X$9

and, more concretely, a sequence of pointed sets

$$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$0

Accordingly, twists of $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$1 are classified up to Brauer action by torsors under the autoequivalence group (Antieau, 2012).

The paper works out complete classifications for genus $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$2 curves, quadrics, and noncommutative projective spaces, and a partial classification for curves of higher genus. In these examples, the twisted Brauer space organizes two kinds of data at once: ordinary Brauer twisting by Azumaya algebras and genuinely new twisting coming from nontrivial torsors under groups of derived autoequivalences.

4. Extended Brauer spaces of spectral algebraic stacks

In spectral algebraic geometry, the relevant object is the extended Brauer space $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$3 of a spectral stack $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$4. Chough defines it as the subspace of invertible objects in the $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$5-category of compactly generated, stable, quasi-coherent stacks on $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$6: $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$7 Its connected components form the extended Brauer group

$$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$8

This definition shifts the emphasis from individual Azumaya algebras to invertible quasi-coherent stacks, though Azumaya algebras still furnish the fundamental examples (Chough, 2020).

The sheafified homotopy groups are computed for quasi-geometric stacks: $$\mathrm{Br}'(X)=H^2(X,\mathbf{G}_m)_{\mathrm{tors}},$$9 For $\mathrm{Br}(X)$0-truncated $\mathrm{Br}(X)$1, this yields

$\mathrm{Br}(X)$2

The appearance of the $\mathrm{Br}(X)$3 term is characteristic of the extended Brauer space and distinguishes it from the more classical cohomological Brauer group (Chough, 2020).

A central theorem states that if $\mathrm{Br}(X)$4 is a quasi-geometric spectral algebraic stack admitting a quasi-finite presentation, then every element of $\mathrm{Br}(X)$5 comes from an Azumaya algebra on $\mathrm{Br}(X)$6. The proof relies on twisted compact generation: for every compactly generated, stable quasi-coherent stack $\mathrm{Br}(X)$7 on $\mathrm{Br}(X)$8, the global section category $\mathrm{Br}(X)$9 is compactly generated. This compact-generation theorem is then used to reconstruct Brauer classes from actual Azumaya objects (Chough, 2020).

This framework gives a higher-categorical version of the principle that “cohomological Brauer classes are geometric.” In particular, the canonical Brauer map is shown to be surjective under the stated hypotheses, extending earlier results of Toën, Antieau–Gepner, and Gabber to quasi-geometric spectral algebraic stacks.

5. Moduli-theoretic interpretations and vanishing phenomena

In moduli theory, the term “Brauer space” is also used in a broader sense for a moduli space parameterizing Brauer classes, often in connection with Azumaya algebras or classifying spaces. Within that perspective, vanishing of the Brauer group is interpreted as the absence of gerbe-theoretic obstructions to universal objects. Biswas, Logares, and Muñoz explicitly state that in the broader context the Brauer space refers to “the moduli space that parameterizes Brauer classes,” and that vanishing means the moduli space acts as a fine moduli space in the gerbe-theoretic sense (Biswas et al., 2010).

Their main theorem concerns the moduli space $\mathbf{PGL}_n$0 of stable pairs with fixed determinant over a smooth projective curve $\mathbf{PGL}_n$1 of genus $\mathbf{PGL}_n$2. If $\mathbf{PGL}_n$3, $\mathbf{PGL}_n$4 has degree $\mathbf{PGL}_n$5, and $\mathbf{PGL}_n$6, then

$\mathbf{PGL}_n$7

The proof combines wall-crossing birational geometry, purity for the Brauer group, projective-bundle exact sequences of Gabber type, and codimension estimates. In this setting, vanishing of $\mathbf{PGL}_n$8 is taken to mean that the moduli of stable pairs carries no nontrivial Brauer obstruction beyond the known projective ambiguity (Biswas et al., 2010).

Related moduli calculations display the same pattern. For the moduli space $\mathbf{PGL}_n$9 of $n$0-stable framed bundles over a smooth projective curve, Biswas, Gómez, and Muñoz show that under the small-parameter assumption

$n$1

one has

$n$2

The argument proceeds by identifying a dense open subset as a projective bundle over the moduli space $n$3 of stable bundles and then applying the exact sequence

$n$4

together with the fact that $n$5 generates $n$6 (Biswas et al., 2011).

By contrast, for the moduli space $n$7 of stable parabolic vector bundles, Biswas and Holla obtain a nontrivial cyclic Brauer group: $n$8 where

$n$9

They further show that the Brauer class of the universal projective bundle restricted to a point of the curve is a generator, and that a universal vector bundle exists over $K_0$0 if and only if $K_0$1 (Biswas et al., 2010).

These examples make precise a common geometric reading of Brauer spaces: triviality corresponds to the disappearance of twisting obstructions, whereas nontrivial cyclic Brauer groups record precisely how universal families fail to exist.

6. Homogeneous spaces, tame stacks, and computational frameworks

Homogeneous spaces provide a complementary testing ground. Borovoi proves that if $K_0$2 is an algebraically closed field of characteristic $K_0$3, $K_0$4 is a connected linear algebraic group, and $K_0$5 is a connected closed subgroup, then

$K_0$6

For smooth proper compactifications, the unramified Brauer group coincides with the cohomological Brauer group, so this is a vanishing theorem for the Brauer-theoretic obstruction on $K_0$7 (Borovoi, 2012).

A 2026 extension considers complex homogeneous spaces $K_0$8 with $K_0$9 connected, simply connected, semisimple and $K$0 closed and connected. In that setting the algebraic Brauer group, the analytic Brauer group, and their cohomological analogs all coincide, and one has the explicit computation

$K$1

If $K$2 is semisimple, this becomes

$K$3

This identifies the Brauer group directly with central extension data of the stabilizer and shows that the algebraic and analytic Brauer spaces agree in this class of examples (Bhaumik et al., 21 May 2026).

For tame algebraic stacks, recent work shifts attention to the relation between a stack and its coarse space. Let $K$4 be the coarse moduli morphism of a tame algebraic stack. Under a locally Brauerless hypothesis one has

$K$5

hence a low-degree exact sequence

$K$6

The stalk of $K$7 at a geometric point is essentially the character group $K$8 of the stabilizer $K$9. Thus the difference between the Brauer group of the stack and that of its coarse space is governed by the cohomology of fiberwise Picard data. The same framework yields formulas for root stacks and computations such as

$\mathrm{Br}(R)$00

for the moduli stack of elliptic curves over a regular noetherian $\mathrm{Br}(R)$01-scheme $\mathrm{Br}(R)$02 (Achenjang, 2024).

Taken together, these results indicate a broad structural picture. Brauer spaces are computed by combining Azumaya representability, descent, and compact generation on the one hand, with Picard-theoretic and stabilizer-theoretic data on the other. Vanishing theorems, exact sequences, and explicit extension formulas are the principal mechanisms by which the geometry of a space, stack, or derived category is translated into Brauer-theoretic information.