Dwyer-Kan Localization in Higher Categories

• Dwyer-Kan localization is a higher-categorical process that inverts selected morphisms while preserving all homotopical coherence and enriched mapping spaces.
• It employs models like simplicial and hammock localization to transition from a 1-categorical framework to a fully ∞-categorical context with universal properties.
• Its broad applications in C*-algebras, operads, props, and Lie–Rinehart pairs unify diverse homotopical methods under a single robust framework.

Dwyer-Kan localization is a foundational construction in higher category theory and homotopy theory that systematically inverts a class of morphisms in a category while preserving all homotopical data, including higher coherences. This process realizes the passage from a 1-categorical context—where morphisms are simply inverted formally—to a fully homotopical or ∞-categorical context, preserving mapping spaces and enriching the traditional homotopy category. Dwyer-Kan localization has far-reaching applications in the study of C*-algebras, operads, coalgebras, props, and Lie–Rinehart pairs, serving as a unifying technique in modern homotopical and categorical frameworks.

1. Universal Construction of Dwyer–Kan Localization

Given a category $C$ and a distinguished class of morphisms $W$ (often weak equivalences), the Dwyer–Kan localization $L_W(C)$ is constructed as an ∞-category (quasi-category or simplicially enriched category) together with a functor

$\ell\colon C \to L_W(C)$

satisfying two conditions:

• $\ell$ sends each $w \in W$ to an equivalence in $L_W(C)$.
• For any quasi-category $D$, pre-composition with $\ell$ induces an equivalence of mapping spaces:

$$\operatorname{Map}\big(L_W(C), D\big) \simeq \operatorname{Map}_W(C, D)$$

where $\operatorname{Map}_W(C, D)$ is the full subcategory of functors $F\colon C \to D$ that send every $w \in W$ to an equivalence in $D$.

At the level of homotopy categories, $L_W(C)$ is initial among all categories under $C$ in which all maps in $W$ become invertible; this recovers the Gabriel–Zisman localization but enhances it to retain higher homotopical data (Meyer, 29 Aug 2025, Hinich, 2013).

Several equivalent explicit models for $L_W(C)$ exist: the simplicial localization, the hammock localization, and (in the ∞-categorical context) the marked nerve construction. In the classical hammock localization, mapping spaces are constructed from diagrams (hammocks or zigzags) that alternate between "inverted" and ordinary morphisms, encoding all higher homotopies (Basterra et al., 2016, Yalin, 2013).

2. Model Structures and Mapping Spaces

When $C$ arises as the underlying category of a (combinatorial, simplicial) model category $(M, W)$, Dwyer–Kan localization $L_{DK}(M)$ enhances the homotopy category $\mathrm{Ho}(M) = M[W^{-1}]$ to an ∞-category whose mapping spaces encode all zigzags of weak equivalences between cofibrant-fibrant objects. Mapping spaces can be made explicit: for cofibrant-fibrant $X,Y$,

$$\operatorname{Map}_{L_{DK}(M)}(X, Y) \simeq N(\mathrm{Weq}(M)_{X/}^{Y/})$$

where $\mathrm{Weq}(M)_{X/}^{Y/}$ is the category of diagrams

$$X \xleftarrow{\sim} X_0 \to Y_0 \xleftarrow{\sim} Y$$

with backward arrows in $W$ (Péroux, 2020, Hinich, 2013). More generally, mapping spaces can be computed via cosimplicial or simplicial resolutions.

In a category of fibrant objects (à la Brown), mapping spaces in the Dwyer–Kan localization reduce to spaces of strict morphisms into path-objects, simplifying calculations in practice (Pištalo, 6 Jan 2026).

3. Functoriality, Equivalences, and Universal Properties

Dwyer–Kan localization is functorial: a functor $F\colon (C, W)\to (D, V)$ sending $W$ into $V$ induces a functor $L_W(C)\to L_V(D)$ between their localizations (Basterra et al., 2016). Moreover, localization enjoys a universal property: for any ∞-category $D$, restriction along $C\to L_W(C)$ induces an equivalence

$$\mathrm{Fun}_W(C, D) \simeq \mathrm{Fun}(L_W(C), D)$$

where $\mathrm{Fun}_W(C, D)$ are those functors carrying $W$ into equivalences (Pištalo, 6 Jan 2026).

A functor between simplicial categories is a Dwyer–Kan equivalence if it induces equivalences of mapping spaces and an equivalence on homotopy categories. Model-categorical Quillen equivalences induce Dwyer–Kan equivalences on localized categories (Yalin, 2013). Weak monoidal Quillen equivalences lift to symmetric monoidal equivalences of Dwyer–Kan localized ∞-categories, ensuring invariance of the homotopy theory under suitable base changes (Péroux, 2020).

4. Applications: C*-Algebras, Operads, Props, and Lie–Rinehart Pairs

C*-Algebras and Bicategory of Correspondences

In the category of (σ-unital) C*-algebras and nondegenerate -homomorphisms, Dwyer–Kan localization at the class of “corner embeddings” $i_H\colon B\to K(H\oplus B)$ produces the bicategory of proper C-correspondences, Corr$^{pr}$. This bicategory has:

• Objects: C*-algebras $A$,
• 1-morphisms: right Hilbert $B$-modules with an appropriate left $A$-action,
• 2-morphisms: unitary isomorphisms of correspondences.

This result provides a powerful universal characterization: any C*-stable, homotopy-theoretic invariant factors uniquely through Corr$^{pr}$, and all higher homotopies collapse to at most 2-cells. Unifying various C*-algebraic constructions (Morita equivalence, Pimsner algebras, crossed products) under a universal localization perspective (Meyer, 29 Aug 2025).

Colored Operads, Props, and Algebraic Theories

The Dwyer–Kan localization controls the ∞-category of operads in a symmetric monoidal model category $S$, where Dwyer–Kan equivalences are maps that are levelwise weak equivalences and induce essentially surjective maps on homotopy categories. Dwyer–Kan localization of operads and props ensures independence of the homotopy theory of algebras (e.g., homotopy invariance under weak equivalence of cofibrant props) and robustly encodes all higher mapping data (Truong, 2023, Yalin, 2013).

Tree-hammock localization extends the Dwyer–Kan formalism to operads and inverts specifically designated arities (e.g., unary operations), with associated universal properties for categories of algebras (Basterra et al., 2016).

Lie–Rinehart Pairs and Homotopy Lie Algebroids

Dwyer–Kan localization of the category of dg Lie–Rinehart pairs $(A, M)$ (with $A$ a semi-free commutative dg-algebra and $M$ a cell $A$-module) at quasi-isomorphisms is equivalent to the localization for strong homotopy Lie–Rinehart pairs. This provides a fully coherent ∞-category, which is the category of fibrant objects for suitable (semi-)model structures. In the finite type case, the Dwyer–Kan localization fibers as a Cartesian fibration over the base cdga, with fibers presentable, and all "BV-type" resolutions are unique up to contractible homotopy (Pištalo, 6 Jan 2026).

5. Symmetric Monoidal and Operadic Structures

Dwyer–Kan localization is compatible with symmetric monoidal structures. For a symmetric monoidal model category $(M, \otimes)$ satisfying the appropriate axioms, $L_{DK}(M)^{\otimes}$ is a symmetric monoidal ∞-category, and weak monoidal Quillen equivalences induce equivalences of such structures. This underpins equivalences such as the Dold–Kan correspondence for coalgebras in the ∞-category setting, and enables systematic transport of operadic and coalgebraic invariants across models (Péroux, 2020, Truong, 2023).

Fiber sequences in the Dwyer–Kan localizations relate mapping spaces of operads, bimodules, and algebra objects, reflecting the expected patterns from enriched category theory. The invariance of these structures under model-theoretic base change and Quillen equivalence is a general feature (Truong, 2023).

6. Limitations and Rigidification Phenomena

While Dwyer–Kan localization robustly generates homotopy-coherent ∞-categories, rigidification fails in nontrivial examples, particularly for coalgebras. For instance, there is no Quillen equivalence between the strict model categories of coalgebras in simplicial modules and in chain complexes, even though their $\infty$-categories of (homotopy-coherent) coalgebras are equivalent via Dwyer–Kan localization. Explicitly, various $\mathrm{A}_\infty$-coalgebras in module spectra have no strict model, in contrast to well-known rigidification theorems for homotopy algebras (Péroux, 2020).

A plausible implication is that, in contrast to the algebraic setting, additional obstructions occur for coalgebraic structures, necessitating strong finiteness or cartesian hypotheses to recover strictification in the coalgebra world.

7. Structural Features and Examples

The Dwyer–Kan localization framework covers a wide range of algebraic, categorical, and topological structures:

Context	Underlying Category	Class W	Target DK Localization
C*-algebras	C*cat$_+$	Corner embeddings	Bicategory of proper correspondences
Operads	Op$(\mathcal{S})$	DK equivalences	∞-cat of operads in $\mathcal{S}$
Lie–Rinehart pairs	dgLR$(k)^{cof}$	Quasi-isomorphisms	∞-cat of SH Lie–Rinehart pairs
Model categories	$(M,W)$	Weak equivalences	$\infty$-cat encoding model structure
Algebras over props	Alg$_P$	Quasi-isomorphisms	Homotopy P-algebras

For each, explicit presentations of mapping spaces and equivalence criteria are available, with universal properties, functoriality, and compatibility with (co)fibrations and base changes established across contexts (Meyer, 29 Aug 2025, Péroux, 2020, Truong, 2023, Basterra et al., 2016, Hinich, 2013, Yalin, 2013, Pištalo, 6 Jan 2026). Uniqueness of certain resolutions (e.g., BV-type in the Lie–Rinehart context) is detected as contractibility of relevant mapping spaces in the Dwyer–Kan localization (Pištalo, 6 Jan 2026).

Dwyer–Kan localization consequently serves as the organizing higher-categorical principle that unifies various flavors of homotopical localization, making it fundamental to modern approaches in algebraic topology, noncommutative geometry, and higher category theory.