Brown–Adams Representability

• Brown–Adams representability is a framework that precisely defines when cohomological functors in a triangulated or stable ∞-category are representable by objects.
• It employs key categorical concepts such as t-structure approximability, strong generation, and purity to ensure that both objects and natural transformations are uniquely determined.
• The theory underpins advances in algebraic geometry, stable homotopy theory, and representation theory by offering practical criteria for functor representability.

Brown–Adams representability is a cornerstone in the theory of triangulated and higher categories, providing precise conditions under which (co)homological functors are representable by actual objects. Originating as a refinement and unification of results by Brown (in algebraic topology) and Adams (in stable homotopy), the Brown–Adams paradigm underpins substantial advances across algebraic geometry, representation theory, and homotopy theory. The contemporary formulation identifies categorical and homological finiteness conditions—expressed via t-structure approximability, strong generation, and purity—as central to establishing when contravariant (or covariant) functors admit representing objects and when natural transformations are induced by morphisms.

1. Formal Definitions and the Brown–Adams Paradigm

Let $\mathcal{T}$ be a compactly generated triangulated (or stable $\infty$-) category with a full subcategory of compact objects $\mathcal{T}^c$. The restricted Yoneda functor

$$y\colon \mathcal{T} \to \mathrm{Flat}(\mathcal{T}^c), \qquad y(X) = \mathrm{Hom}_\mathcal{T}(-, X)|_{\mathcal{T}^c}$$

identifies the class of cohomological functors on $\mathcal{T}^c$ with flat modules over the additive category $\mathcal{T}^c$. Brown–Adams representability posits two core properties:

• Essential Surjectivity: Every cohomological functor $H : (\mathcal{T}^c)^{op} \to \mathrm{Ab}$ arises as $$H \cong \mathrm{Hom}_\mathcal{T}(-, X)|_{\mathcal{T}^c}$$ for some $X \in \mathcal{T}$.
• Fullness (Morphisms): Every natural transformation between such functors is induced by a unique morphism $X \to Y$ in $\mathcal{T}$.

A category $\mathcal{T}$ is termed a Brown category if both properties hold; this is frequently called Brown–Adams representability, especially when emphasizing the role of morphisms as in Adams' original formulation (Bird, 14 Jan 2026, Rubinstein, 2024).

2. Categorical Axiomatization: Approximability and Purity

Modern Brown–Adams results require subtle categorical hypotheses, generalizing classical compact (or well-) generation:

• Approximable Triangulated Categories: An $\mathbb{E}$-linear triangulated category with coproducts is approximable if it has a single compact generator $G$ and there exists an integer $n$ so that every object $X$ admits a triangle with objects from the preferred aisle/co-aisle associated to a $t$-structure generated by $G$, and with shifts constrained to a finite window (Neeman, 2018).
• Purity: Purity is encoded in the global pure projective (or injective) dimension of $\mathrm{Flat}(\mathcal{T}^c)$, arising via the restricted Yoneda embedding. The vanishing of higher pure $\mathrm{Ext}$ enforces that all cohomological functors—and all morphisms between them—are detected by representables and honest morphisms (Bird, 14 Jan 2026, Muro et al., 2013).

These structural hypotheses replace assumptions such as preservation of all coproducts, providing a finer invariant that captures more general and geometric settings.

3. Brown–Adams Theorems: Principal Statements and Proof Strategies

The main contemporary Brown–Adams theorem asserts:

• If $\mathcal{T}$ is compactly generated and $\mathcal{T}^c$ is idempotent complete, strongly generated, and small (typically countable), and if $$\mathrm{Ext}^i_{\mathrm{Mod}(\mathcal{T}^c)}(F, G) = 0$$ for all $i \geq 2$ and all flat $F, G$, then:
  1. Every cohomological functor $F: (\mathcal{T}^c)^{op} \to \mathrm{Ab}$ arises from a unique object $X\in\mathcal{T}$.
  2. Every natural transformation between such functors is induced by a map in $\mathcal{T}$ (Bird, 14 Jan 2026, Rubinstein, 2024).

• In the context of triangulated categories with a graded ring action, a strongly generated, Ext-finite, idempotent complete $R$-linear category has the property that a graded $R$-linear, cohomological functor is representable exactly when it is locally finite (takes values in finitely generated $R$-modules) (Letz, 2022).

Proofs rely centrally on constructing approximation towers via homotopy colimits, controlling vanishing of so-called phantom maps, and applying obstruction theory for extending finite-length truncated Postnikov towers (Neeman, 2018, Muro et al., 2013).

4. Key Variants, Duality, and Limitations

Brown–Adams representability extends to several dual or parameterized variants:

• Dual Brown Representability: In homological contexts, representability for covariant, product-preserving, homological functors may fail unless the base abelian category has a product generator (Modoi, 2012). The analogy with stable homotopy is only valid when compactness for products and coproducts coincide, which is not true in homotopy categories of complexes for arbitrary rings (Modoi et al., 2010).
• Transfinite Adams Representability: For well-generated triangulated categories and arbitrary regular cardinals $\kappa$, the restricted Yoneda functor $S_\kappa$ (to $\kappa$-products) admits representability iff the projective dimension of every flat functor is at most one. This generalizes classical Adams representability and quantitatively links to the (pure-)global dimension of the underlying abelian category (Muro et al., 2013, Bird, 14 Jan 2026).
• Obstructions and Failure: Brown–Adams representability fails in non-well-generated categories, e.g., $K(\mathrm{Ab})$; existence of all coproducts or products is insufficient (Modoi et al., 2010). Pure global dimension at least two in the underlying module category prevents Brown–Adams representability in derived categories (Bird, 14 Jan 2026).

5. Applications, Concrete Criteria, and Transfer

The paradigmatic applications of Brown–Adams representability include:

• Algebraic Geometry: For derived categories of quasi-compact, separated schemes, approximability and finiteness of cohomology ensure that cohomological functors on perfect complexes are representable, yielding the classical Bondal–Van den Bergh theorem and GAGA-adjoints (Neeman, 2018, Modoi, 2024).
• Stable Homotopy Theory: The homotopy category of spectra is approximable, and Brown–Adams gives that every cohomological functor on finite spectra is represented by a spectrum, fully capturing Adams' theorem (Neeman, 2018, Rubinstein, 2024).
• Representation Theory: In the presence of a graded Noetherian ring acting on a strongly generated triangulated category (e.g., bounded derived categories of group rings), every locally finite graded cohomological functor is representable. Group cohomology, complete intersections, and Hochschild cohomology all fall into this exact paradigm (Letz, 2022).
• Motivic and Tensor-Triangulated Contexts: Brown–Adams methods are essential in transporting stratification results between derived categories of motivic and representation theoretic origin, relying on identification of compacts and residue fields at points of the spectrum (Rubinstein, 2024).

Transfer results established via definable functors show that if a functor between compactly generated triangulated categories is fully faithful and definable, Brown–Adams representability transfers in both directions (Bird, 14 Jan 2026).

6. The Role of Purity, Definable Functors, and Obstruction Theory

Purity, as measured by the pure global dimension of $\mathrm{Flat}(\mathcal{T}^c)$, is critical:

• Brown–Adams representability is governed by the vanishing of $\mathrm{PExt}^i$ for $i \geq 2$; categories with higher pure global dimension fail representability (Bird, 14 Jan 2026).
• Definable functors preserve purity and allow the transfer of representability properties under base change, localization, and passage to subcategories (e.g., from $\mathcal{D}_{qc}(U)$ to $\mathcal{D}_{qc}(X)$ in schemes) (Bird, 14 Jan 2026).
• Obstruction theory, built via Postnikov resolutions, links the existence and uniqueness of extensions of truncated systems to vanishing $\mathrm{Ext}$ groups in the module category over compacts. The Adams spectral sequence differentials directly calculate these obstructions (Muro et al., 2013).

7. Illustrative Examples and Corollaries

Category/Class	Sufficient Criterion for Brown–Adams	Source
Derived category of von Neumann regular ring $R$	$\mathrm{pgldim}(R) \leq 1$	(Bird, 14 Jan 2026)
Derived category $\mathcal{D}_{qc}(X)$, $X$ scheme	$\mathrm{pgldim}(\mathcal{O}_{X,x}) \leq 1$ for all $x$	(Bird, 14 Jan 2026)
Stable $\infty$-category	Compact generation, small compacts	(Nguyen et al., 2021)
Homotopy category $K(\mathrm{Ab})$	No representability: not well-generated/compactly generated	(Modoi et al., 2010)

Examples demonstrate that for $R$ von Neumann regular, the derived category satisfies the telescope conjecture, Freyd's generating hypothesis, and Brown–Adams representability. For arbitrary schemes, failure of regularity at any point precludes representability for the entire derived category. In motivic and homotopical settings, stratification and object generation via residue fields require Brown–Adams as a technical tool for lifting structure.

---

Brown–Adams representability offers a unified, categorical framework for understanding when and how cohomological functors—or morphisms between such functors—arise from and correspond to actual objects and morphisms within triangulated or stable homotopical contexts. Its realization depends on precise and computable invariants such as (pure) global dimension, compact generation, and the interplay of definable functors and categorical purity. The theory encapsulates and extends foundational results in homotopy theory, algebraic geometry, and representation theory, while also diagnosing sharp limitations and obstructions in their failure cases (Neeman, 2018, Letz, 2022, Nguyen et al., 2021, Modoi, 2024, Muro et al., 2013, Modoi et al., 2010, Modoi, 2012, Rubinstein, 2024, Bird, 14 Jan 2026).