Continuous Six-Functor Formalisms

• Continuous six-functor formalisms are a framework integrating six classical operations with additional descent and continuity conditions, enabling consistent behavior over gluing, limits, and colimits.
• They characterize sheaf theories on locally compact Hausdorff spaces by demonstrating that every continuous formalism is equivalent to a sheaf formalism with coefficients in the fiber over a point.
• Applications span derived, analytic, and motivic settings, where these formalisms facilitate duality principles, smoothness conditions, and cohomological descent in higher category theory.

Continuous six-functor formalisms are six-functor formalisms equipped with additional descent and continuity conditions controlling how the fibers $D(X)$ vary with the underlying space or stack and how the operations $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$ interact with gluing, limits, and colimits. In the locally compact Hausdorff setting, the central model is the assignment $X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$, and a principal structural result is that this formalism is initial among all continuous six-functor formalisms valued in dualizable presentable stable $\infty$-categories (Zhu, 17 Jul 2025). A subsequent characterization shows that every continuous six-functor formalism on $\mathrm{LCH}$ is equivalent to $\mathrm{Shv}(-,D(\mathrm{pt}))$, so the continuous theory on $\mathrm{LCH}$ is, in a precise sense, exhausted by sheaf formalisms with coefficients in the value at a point (Bunke et al., 9 Jun 2026).

1. Historical and conceptual background

The modern subject sits at the intersection of Grothendieck-style operations, higher category theory, and descent. In the derivator setting, a derivator six-functor-formalism on a category $\mathcal S$ is defined as a strict morphism of pre-2-multiderivators

$p:D\to S^{\cor}$

that is simultaneously a left fibered multiderivator and a right fibered multiderivator, thereby encoding not only the six functors but also their interaction with homotopy Kan extensions (Hörmann, 2017). The same framework emphasizes that one effectively obtains a “nine-functor-formalism,” since the adjoint structure and derivator enhancement internalize additional companions needed for descent and coherence (Hörmann, 2017).

A subsequent construction shows that derivator six-functor-formalisms can be built from compactifications starting from a derivator-enhanced four-functor formalism satisfying base-change and projection-formula axioms (Hörmann, 2019). In that setting, the formalism is designed to encode all isomorphisms between compositions of the six functors and their compatibilities, and it is explicitly intended to extend to stacks using cohomological descent (Hörmann, 2019).

This earlier derivator work already isolates two themes that remain decisive in later “continuous” versions: first, the six operations are best organized on categories of correspondences rather than on ordinary morphisms alone; second, descent is not an accessory property but part of the definition of a usable formalism. Later work on locally compact Hausdorff spaces, rigid spaces, diamonds, and condensed or analytic contexts refines these ideas inside presentable stable $\infty$-categories.

2. Abstract structure of a six-functor formalism

On locally compact Hausdorff spaces one works with the Nagata context

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$0

and a six-functor formalism is expressed by a functor

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$1

or, equivalently in another formulation, as a functor

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$2

with each $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$3 a closed symmetric monoidal stable $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$4-category and each $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$5 inducing a colimit-preserving pullback $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$6 with right adjoint $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$7 (Zhu, 17 Jul 2025, Bunke et al., 9 Jun 2026). Open embeddings $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$8 are required to admit left adjoints $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$9, proper maps $X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$0 are required to have $X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$1 preserve colimits and hence admit the exceptional right adjoint $X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$2, and the system must satisfy the standard base-change, projection-formula, and mixed Beck–Chevalley compatibilities (Bunke et al., 9 Jun 2026).

In this formalism the six operations are

$X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$3

For open and proper morphisms the characteristic compatibilities include

$X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$4

and for a Cartesian square with $X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$5 and $X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$6,

$X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$7

These identities are the higher-categorical form of the usual projection and exchange laws (Bunke et al., 9 Jun 2026).

A broader classification framework formulates a six-functor formalism on a geometric setup $X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$8 as a lax symmetric-monoidal functor

$X\mapsto \mathrm{Shv}(X;\mathrm{Sp})$9

with adjoint pairs $\infty$0 and $\infty$1, together with closed monoidal structures (Nabaala, 17 Jul 2025). This general language is designed to accommodate algebraic, topological, rigid-analytic, and motivic examples within a single correspondence-based architecture.

3. Continuity on locally compact Hausdorff spaces

The adjective “continuous” acquires a precise meaning on $\infty$2. A continuous six-functor formalism in Zhu’s sense is a Nagata six-functor formalism valued in dualizable presentable stable $\infty$3-categories, with colimit-preserving natural transformations, satisfying three additional axioms: canonical descent, profinite descent, and hyperdescent (Zhu, 17 Jul 2025).

Canonical descent consists of localization and filtered-open continuity. For each closed inclusion with open complement one requires a Cartesian square in $\infty$4, and for an increasing union $\infty$5 one requires

$\infty$6

along restriction functors (Zhu, 17 Jul 2025). Profinite descent requires that if $\infty$7 is a cofiltered limit of compact Hausdorff spaces, then

$\infty$8

with transition maps given by pushforwards (Zhu, 17 Jul 2025). Hyperdescent requires that on hypercomplete spaces, equivalences are detected on stalks; equivalently, the product of the stalk functors $\infty$9 reflects equivalences (Zhu, 17 Jul 2025).

Bunke and Volpe give a closely related axiomatization in terms of canonical descent, section-determinedness, finitariness, and dualizability of $\mathrm{LCH}$0 (Bunke et al., 9 Jun 2026). Here section-determined means that evaluation on opens is jointly conservative, and finitary means that for a cofiltered system of compact Hausdorff spaces with limit $\mathrm{LCH}$1,

$\mathrm{LCH}$2

They prove that continuity in Zhu’s sense implies section-determinedness, so the two formulations are aligned at the level needed for characterization theorems (Bunke et al., 9 Jun 2026).

A recurring point in the literature is that “continuity” is context-dependent. On $\mathrm{LCH}$3 it refers to these descent and point-detection axioms (Zhu, 17 Jul 2025). In rigid-analytic categories of nuclear sheaves, the same word is used for the fact that $\mathrm{LCH}$4 preserve colimits and that pushforwards satisfy base-change and projection formulae (Montagnani, 20 May 2026). On light condensed anima, continuity is expressed by requiring that $\mathrm{LCH}$5 preserve all limits (He, 22 Nov 2025).

4. Sheaf-theoretic realization and universal properties

For spectral sheaves on locally compact Hausdorff spaces, the six operations can be described explicitly. For a continuous map $\mathrm{LCH}$6, the direct image

$\mathrm{LCH}$7

has a left adjoint $\mathrm{LCH}$8 given on presheaves by left Kan extension along $\mathrm{LCH}$9 followed by sheafification (Zhu, 17 Jul 2025). For an open immersion $\mathrm{Shv}(-,D(\mathrm{pt}))$0, one has extension by zero

$\mathrm{Shv}(-,D(\mathrm{pt}))$1

and Verdier duality produces the proper-pushforward and exceptional pullback (Zhu, 17 Jul 2025). These operations satisfy base-change, projection formula, and mixed Beck–Chevalley for open and proper morphisms (Zhu, 17 Jul 2025).

The principal universal statement is that the assignment

$\mathrm{Shv}(-,D(\mathrm{pt}))$2

is the initial object of the $\mathrm{Shv}(-,D(\mathrm{pt}))$3-category $\mathrm{Shv}(-,D(\mathrm{pt}))$4 of continuous six-functor formalisms (Zhu, 17 Jul 2025). Equivalently, for any other continuous formalism $\mathrm{Shv}(-,D(\mathrm{pt}))$5, there is a unique colimit-preserving natural transformation

$\mathrm{Shv}(-,D(\mathrm{pt}))$6

unique up to contractible choice (Zhu, 17 Jul 2025). This isolates spectral sheaves as the universal coefficient system among continuous six-functor theories on $\mathrm{Shv}(-,D(\mathrm{pt}))$7.

The characterization theorem sharpens this result. If a six-functor formalism on $\mathrm{Shv}(-,D(\mathrm{pt}))$8 has canonical descent, is section-determined, has finitary cohomology functor $\mathrm{Shv}(-,D(\mathrm{pt}))$9, and $\mathrm{LCH}$0 is dualizable, then

$\mathrm{LCH}$1

as six-functor formalisms (Bunke et al., 9 Jun 2026). In particular, every continuous six-functor formalism in Zhu’s sense is equivalent to a sheaf formalism with coefficients in the fiber over a point (Bunke et al., 9 Jun 2026). This removes a potential ambiguity: on $\mathrm{LCH}$2, continuous six-functor formalisms are not merely analogous to sheaf theories; they are characterized by sheaf theory.

The same universal perspective has consequences for invariants of coefficient categories. For any finitary continuous localizing invariant

$\mathrm{LCH}$3

and any continuous six-functor formalism $\mathrm{LCH}$4, one has

$\mathrm{LCH}$5

so localizing invariants of continuous six-functor formalisms behave like compactly supported sheaf cohomology theories (Zhu, 17 Jul 2025).

5. Extensions of the continuous formalism beyond spectral sheaves

A major extension on locally compact Hausdorff spaces replaces spectra by an arbitrary closed symmetric monoidal $\mathrm{LCH}$6-category $\mathrm{LCH}$7 that is stable and bicomplete. In that setting, one obtains a self-contained six-functor formalism for $\mathrm{LCH}$8-valued sheaves, including $\mathrm{LCH}$9, without assuming presentability and without restricting to hypercomplete sheaves (Volpe, 2021). The construction uses Lurie’s tensor product of cocomplete $\mathcal S$0-categories, covariant Verdier duality

$\mathcal S$1

the strong dualizability of $\mathcal S$2, and the identification

$\mathcal S$3

for stable bicomplete $\mathcal S$4 (Volpe, 2021). The same work proves localization sequences without presentability and shows that if $\mathcal S$5 induces a locally contractible geometric morphism, then $\mathcal S$6 preserves colimits and is related to $\mathcal S$7 by a smooth-pullback formula (Volpe, 2021).

In analytic geometry, Montagnani studies the assignment

$\mathcal S$8

for derived rigid spaces and proves that it extends to a symmetric-monoidal 6-functor

$\mathcal S$9

with pullback, pushforward, exceptional pullback, exceptional pushforward, relative tensor, and enriched Hom (Montagnani, 20 May 2026). Here $p:D\to S^{\cor}$0, and for any $p:D\to S^{\cor}$1, the functors $p:D\to S^{\cor}$2 all preserve colimits; $p:D\to S^{\cor}$3 and $p:D\to S^{\cor}$4 satisfy base-change and projection formula; and pullbacks carry nuclear objects to nuclear objects (Montagnani, 20 May 2026). This is a genuinely continuous six-functor theory in the sense relevant to presentable rigid-analytic coefficient categories.

For diamonds and $p:D\to S^{\cor}$5-stacks, a six-functor formalism is constructed for nuclear $p:D\to S^{\cor}$6-sheaves, where $p:D\to S^{\cor}$7 is a nuclear $p:D\to S^{\cor}$8-algebra and prominent cases include $p:D\to S^{\cor}$9, $\infty$0, and $\infty$1 (Mann, 2022). The categories $\infty$2 are presentable symmetric monoidal stable $\infty$3-categories, $\infty$4 is exact and preserves all colimits, $\infty$5 exists for $\infty$6-fine morphisms and preserves colimits, $\infty$7 is right adjoint to $\infty$8, and the formalism satisfies functoriality, projection formula, proper base-change, and Verdier duality (Mann, 2022). The same paper emphasizes that a genuine six-functor formalism had been missing for non-torsion $\infty$9-adic coefficients and that the nuclear formalism works integrally at the $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$00-level and then inverts $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$01 to reach $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$02 or $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$03 (Mann, 2022).

A parallel $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$04-adic theory in rigid-analytic geometry constructs the stable symmetric-monoidal $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$05-categories

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$06

on small $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$07-stacks and develops $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$08, together with proper base-change, projection formula, and duality (Mann, 2022). The construction depends on condensed mathematics, solid almost modules, and descendable maps, and it is used to prove $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$09-torsion Poincaré duality and a $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$10-torsion Riemann–Hilbert correspondence (Mann, 2022).

The domain of universal continuous formalisms has also been enlarged from $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$11 to light condensed anima. In that setting, the extension of

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$12

is initial among six-functor formalisms satisfying dualizability and point-conservativity on light profinite sets together with continuity on condensed anima (He, 22 Nov 2025). This places Zhu’s universal result inside a broader condensed-geometric setting.

6. Duality, smoothness, and major applications

One of the central uses of continuous six-functor formalisms is the organization of duality. For sheaves on locally compact Hausdorff spaces with coefficients in a stable bicomplete closed symmetric monoidal $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$13-category, the formalism supports covariant Verdier duality and expresses Atiyah duality internally (Volpe, 2021). If $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$14 is a proper submersion of smooth manifolds with relative tangent bundle $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$15, then

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$16

and in the global case $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$17 for a compact smooth manifold,

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$18

recovering classical Atiyah duality (Volpe, 2021).

For spectral sheaves on locally compact Hausdorff spaces, the formalism provides a language for cohomological smoothness. If $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$19 satisfies a categorical Künneth formula

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$20

then for a map $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$21 the existence of a $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$22-linear left adjoint $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$23 to $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$24, the condition that $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$25 is $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$26-suave, and universal $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$27-local contractibility are equivalent (Land et al., 30 Jun 2026). In this framework the relative dualizing object is $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$28, and if $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$29 is compact, $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$30-smooth of homotopy dimension $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$31, one gets a Poincaré duality equivalence

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$32

together with a Künneth formula for cohomology (Land et al., 30 Jun 2026). The same methods are applied to ANR homology manifolds, Spivak tangent fibrations, and generalizations of Wilder’s monotone mapping theorem (Land et al., 30 Jun 2026).

In rigid-analytic geometry, the six-functor formalism on nuclear sheaves is used to define and study smooth and proper $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$33-categories. A presentable stable $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$34-category $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$35 is smooth over $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$36 if the coevaluation admits a colimit-preserving right adjoint, proper over $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$37 if the evaluation admits a colimit-preserving right adjoint, and fully dualizable if both internally smooth and internally proper (Montagnani, 20 May 2026). For a classical rigid variety $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$38, one has

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$39

and for $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$40 smooth, proper, qcqs, separated over $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$41,

$f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$42

(Montagnani, 20 May 2026). The same paper constructs a smooth, proper, non-algebraizable rigid variety $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$43 for which $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$44 is internally smooth but not atomically generated, giving a counterexample to naive extensions of “smooth $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$45 compactly generated” in the $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$46-linear rigid-analytic setting (Montagnani, 20 May 2026).

The six-functor formalism also continues to serve as the operative background in adjacent theories. In motivic-stable settings such as $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$47, one has $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$48 under the usual geometric hypotheses, together with projection formula, base-change, and purity; these tools are used, for example, in the construction of $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$49-Euler classes, integrality statements, and local index formulas (Bachmann et al., 2020). This suggests that continuous six-functor formalisms on topological and analytic spaces are part of a larger pattern in which descent-compatible duality data organize both cohomological operations and geometric invariants.

A central clarification emerging from the recent literature is therefore twofold. First, continuity is not merely a regularity condition on functors; it encodes localization, descent, and pointwise detectability. Second, on $f^*, f_*, f_!, f^!, \otimes,\underline{\Hom}$50 these conditions are rigid enough that the theory collapses to sheaf theory with coefficients in the fiber over a point (Zhu, 17 Jul 2025, Bunke et al., 9 Jun 2026). In other geometric settings the same paradigm persists, but with different ambient categories, coefficient objects, and continuity conditions.