Benson–Krause–Schwede Canonical Class

• The BKS canonical class is a key invariant in Hochschild cohomology, defined via a ternary cocycle m₃ that reflects obstruction to module realization and A₃-formality.
• Its construction methodically lifts chain-level algebra structures using maps f₁ and f₂, ensuring independence from choices and enabling functorial behavior under differential graded algebra morphisms.
• In practical terms, the canonical class distinguishes whether a graded module is realizable in group cohomology, as seen in studies on Demushkin and generalized quaternion groups.

The Benson–Krause–Schwede (BKS) canonical class is a fundamental invariant in the Hochschild cohomology of group cohomology algebras and related differential graded algebras, introduced in the work of D. Benson, H. Krause, and S. Schwede. It arises as a universal cohomological obstruction related to the realization of module structures, the secondary multiplication structure on cohomology, and $A_3$-formality. The canonical class plays a central role in modern obstruction theory for both finite group cohomology and profinite group cohomology, especially in contexts where explicit higher structure on the cohomology algebra controls deformation and realization problems.

1. Formal Definition and Construction

Let $A$ be a connected differential graded algebra (dga) over a field $F$, with $A^i = 0$ for $i < 0$ and $A^0 = F$, equipped with a differential $\delta$ of degree $+1$. The graded cohomology $H^n(A)$ inherits the cup product $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$. By Kadeishvili's theorem, $A$0 admits a minimal $A$1-structure: maps $A$2, together with an $A$3-quasi-isomorphism $A$4.

The BKS canonical cochain $A$5 is precisely the ternary operation

$A$6

where $A$7 is the multiplication in $A$8. $A$9 is verified to be a Hochschild 3-cocycle of internal degree $F$0 (Lemma 4.3 in (Pál et al., 12 Jan 2026)), yielding the canonical class

$F$1

This class is independent of the choices of $F$2 and $F$3, and is functorial under morphisms of dg-algebras (Proposition 4.5 in (Pál et al., 12 Jan 2026)).

When $F$4 is a group (co)homology dga, such as $F$5 or Tate cohomology $F$6, $F$7 specializes to the canonical class associated to the group.

2. Computation and Module-Theoretic Role

Within Tate or continuous group cohomology, $F$8 is typically taken as $F$9 for a profinite or finite group $A^i = 0$0 and field $A^i = 0$1, and $A^i = 0$2.

The construction of the canonical class, as systematically developed in (0911.3603) and (Pál et al., 12 Jan 2026), utilizes explicit lifts of cocycles and homotopies:

• A $A^i = 0$3-projective resolution $A^i = 0$4 is fixed.
• A chain-level lift $A^i = 0$5 of the algebra product and a secondary homotopy $A^i = 0$6 are constructed, satisfying $A^i = 0$7.
• The ternary cocycle is $A^i = 0$8.

This data is used to build $A^i = 0$9, which is the main object of study for realization problems: $i < 0$0

The BKS theorem asserts that for a graded $i < 0$1-module $i < 0$2, $i < 0$3 acts via

$i < 0$4

and

$i < 0$5

Hence, $i < 0$6 precisely measures the non-realizability of $i < 0$7 as a summand of an actual group cohomology module (0911.3603).

3. Case Studies: Demushkin and Quaternion Groups

Pro-$i < 0$8 Demushkin Groups

For pro-$i < 0$9 Demushkin groups $A^0 = F$0, the canonical class $A^0 = F$1 is analyzed using the Koszul complex $A^0 = F$2 of $A^0 = F$3, which is a quadratic Koszul algebra (Corollary 6.7 in (Pál et al., 12 Jan 2026)). The explicit classification of Koszul basis elements (Lemma 8.2) and the computation of the restriction $A^0 = F$4 allows the identification of $A^0 = F$5 with the coboundary or non-coboundary property of certain cochains.

• For $A^0 = F$6-invariant $A^0 = F$7, one finds that explicit cochains exist killing all $A^0 = F$8 values (Theorem 7.1), leading to $A^0 = F$9 and ensuring $\delta$0-formality.
• For $\delta$1 (notably at $\delta$2, $\delta$3), there remains a tensor (e.g., $\delta$4) where no corresponding coboundary can be found, leading to $\delta$5 (Theorem 7.4).

Generalized Quaternion Groups

For generalized quaternion groups $\delta$6 in characteristic $\delta$7, the Tate cohomology algebra $\delta$8 is 4-periodic, and the canonical cocycle $\delta$9 yields values such as $+1$0 (0911.3603). For $+1$1 ($+1$2), there exists a module $+1$3 with $+1$4, yielding a classical non-realizable module (Section 4 in (0911.3603)). For $+1$5, despite $+1$6 in $+1$7, it acts trivially on all modules, so every graded module is realizable.

4. Connection to $+1$8-formality and Obstructions

$+1$9-formality is the property of a dga that its minimal $H^n(A)$0-model (with differentiated higher multiplication $H^n(A)$1) is quasi-isomorphic to its cohomology algebra equipped only with $H^n(A)$2, i.e., all higher Massey products vanish universally. According to Kadeishvili's obstruction theory (Theorem 5.1 in (Pál et al., 12 Jan 2026)), the single obstruction to $H^n(A)$3-formality is the BKS canonical class: $H^n(A)$4

For Demushkin groups with $H^n(A)$5, $H^n(A)$6 is $H^n(A)$7-formal; for $H^n(A)$8 ($H^n(A)$9, $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$0), it is not. All triple Massey products vanish for Demushkin groups, but the nontrivial $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$1 revealed by the canonical class persists whenever $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$2.

In the module-theoretic context, the BKS class measures module realizability: its vanishing (or not) on a module is equivalent to that module splitting off from an actual group cohomology module (0911.3603).

5. Functoriality, Broader Implications, and Open Questions

The canonical class is functorial under dga (and group cohomology) morphisms, suggesting the possibility of an obstruction theory for larger classes of (profinite) groups. The methods employed for Demushkin groups extend to "elementary type" pro-$m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$3-groups, with the functoriality of $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$4 providing a tool for tracking higher structure across group extensions (Pál et al., 12 Jan 2026). Furthermore, evidence suggests all local and global Galois groups $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$5 at odd $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$6 may have $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$7, raising questions about the extent of $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$8-formality within arithmetic.

More generally, the theory predicts higher $m_2 : H^p(A) \otimes H^q(A) \to H^{p+q}(A)$9-obstruction classes in $A$00, and the vanishing of these classes has implications for rigidity and deformation theory in Galois cohomology (see remarks referencing Positselski's work in (Pál et al., 12 Jan 2026)). For $A$01 or small $A$02, such higher obstructions are expected to arise.

6. Summary Table: Properties of the Canonical Class in Key Contexts

Context	$A$03 in $A$04	Realizability Criterion
Demushkin, $A$05 ($A$06 odd)	$A$07	All modules realizable
Demushkin, $A$08 ($A$09)	$A$10	Obstruction to $A$11-formality
Quaternion, $A$12 (char 2)	$A$13	Non-realizable modules exist
Quaternion, $A$14, $A$15 (char 2)	$A$16	All modules realizable

For Demushkin groups, Koszul property plus vanishing triple Massey products do not suffice for $A$17-formality, as exhibited by the persistence of $A$18 at $A$19. For finite groups, the canonical class provides a single cohomological invariant regulating the entire realization problem for graded modules.

7. Literature and Foundational Results

The original construction and realization criterion for the canonical class are established in Benson, Krause, and Schwede (see (0911.3603)). The explicit computational and obstruction-theoretic role in $A$20-formality, particularly for pro-$A$21 Demushkin groups, is developed in Pál–Quick (Pál et al., 12 Jan 2026). The interplay between the canonical class, Massey products, Koszul algebras, and realization problems reflects current understanding at the interface of group cohomology, homotopical algebra, and algebraic deformation theory.