Weak Monoidal Quillen Pair

• Weak monoidal Quillen pairs are Quillen adjunctions between monoidal model categories that preserve the tensor structure up to weak equivalence on cofibrant objects.
• They use lax comonoidal or lax symmetric monoidal structures where comparison maps become weak equivalences under cofibrancy hypotheses.
• These pairs facilitate the transfer of operadic and algebraic structures and induce genuine monoidal equivalences after localization.

Searching arXiv for recent and foundational papers on weak monoidal Quillen pairs and closely related monoidal Quillen descent results. A weak monoidal Quillen pair is a Quillen adjunction between monoidal model categories in which the monoidal structure is preserved not strictly, but up to weak equivalence on the homotopically relevant subcategory of cofibrant objects. In the formulation used by Schwede–Shipley and adopted in later work, the left adjoint carries a colax, or lax comonoidal, structure, and the associated tensor and unit comparison maps are required to be weak equivalences under cofibrancy hypotheses (Péroux, 2020). This notion occupies a central position in homotopical algebra because it identifies the amount of multiplicative compatibility needed for a Quillen adjunction to induce a genuine monoidal equivalence after passage to Dwyer–Kan localization or to the associated symmetric monoidal $\infty$-categories (Péroux, 2020). Related work also shows that monoidal and two-variable Quillen structures descend canonically to localizations even when the exact classical phrase is absent from the formalism, thereby clarifying the homotopy-invariant content of weak monoidality (Mazel-Gee, 2015).

1. Definition and formal structure

A weak monoidal Quillen pair is defined for symmetric monoidal model categories

$$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$

together with a Quillen adjunction

$\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$

In the formulation recorded in "Coalgebras in the Dwyer-Kan localization of a model category" (Péroux, 2020), the left adjoint $L$ is lax comonoidal, and two conditions are imposed. First, for cofibrant $X,Y\in C$, the comparison map

$L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$

must be a weak equivalence in $D$. Second, for some, hence any, cofibrant replacement

$\lambda:c\mathbb I \xrightarrow{\sim} \mathbb I$

in $C$, the composite

$\begin{tikzcd} L(c\mathbb I)\ar{r}{L(\lambda)} & L(\mathbb I)\ar{r} & \mathbb J \end{tikzcd}$

must be a weak equivalence in $$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$0 (Péroux, 2020). A weak monoidal Quillen equivalence is a weak monoidal Quillen pair whose underlying Quillen adjunction is also a Quillen equivalence (Péroux, 2020).

A parallel formulation appears in "Equivalence of Operads over Symmetric Monoidal Categories" (Nguemo, 2018), where the structure is described on the right adjoint side. There, $$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$1 is required to be lax symmetric monoidal by means of maps

$$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$2

and the adjoint comparison

$$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$3

must be a weak equivalence for cofibrant $$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$4, while the unit comparison

$$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$5

must also be a weak equivalence (Nguemo, 2018). These left- and right-adjoint formulations encode the same homotopical principle: monoidal preservation is only required up to weak equivalence, and only where cofibrancy ensures derived control.

This weakening is both structural and homotopical. Structural, because strict monoidal isomorphisms are replaced by colax or lax comparison morphisms. Homotopical, because those morphisms need only be weak equivalences, and only on cofibrant inputs (Péroux, 2020). This suggests that weak monoidal Quillen pairs are designed to capture derived multiplicativity rather than strict monoidal functoriality.

2. Comparison with stronger and weaker notions

The weak monoidal Quillen condition lies between strict monoidal Quillen adjunctions and broader relative-categorical comparisons of strict and lax morphisms. A stronger condition would require actual monoidal isomorphisms

$$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$6

or equivalently a strong monoidal left adjoint. Several papers in the surrounding literature work with this stronger form rather than the weak one. For example, "Smith Ideals of Operadic Algebras in Monoidal Model Categories" studies a cokernel–kernel adjunction whose left adjoint is strong symmetric monoidal and whose right adjoint is lax symmetric monoidal, then proves Quillen equivalence statements for operadic algebra categories (White et al., 2017). Likewise, "Admissible replacements for simplicial monoidal model categories" explicitly distinguishes weak monoidal and strong monoidal Quillen equivalences, but states that all monoidal Quillen equivalences used there are strong monoidal Quillen equivalences (Bayındır et al., 2020).

At the other end, some work is relevant only by analogy. "Extending homotopy theories across adjunctions" does not study model categories or weak monoidal Quillen pairs, but instead proves that under suitable hypotheses one can enlarge a category of strict monoidal maps to lax, oplax, or pseudo maps without changing the homotopy theory (Gurski et al., 2015). This does not supply tensor comparison maps of Quillen type, but it clarifies a nearby phenomenon: homotopical invariance under weakening of multiplicative morphism notions.

The notion also differs from the $$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$7-categorical descent formalism developed in "Model $$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$8-categories II: Quillen adjunctions" (Mazel-Gee, 2015). That paper does not define weak monoidal Quillen pairs in the classical Schwede–Shipley sense and does not discuss comparison morphisms such as

$$(C,\otimes,\mathbb I) \qquad\text{and}\qquad (D,\wedge,\mathbb J)$$9

Instead, it proves that Quillen adjunctions descend to adjunctions on localizations, that two-variable Quillen adjunctions descend to derived two-variable adjunctions, and that monoidal model $\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$0-categories descend to monoidal localizations (Mazel-Gee, 2015). A plausible implication is that weak monoidal Quillen pairs should be understood as one model-categorical mechanism for producing the sort of monoidal descent that becomes intrinsic at the $\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$1-categorical level.

3. Two-variable Quillen structure and monoidal compatibility

The formal substrate of weak monoidality is the theory of two-variable Quillen adjunctions. In "Model $\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$2-categories II: Quillen adjunctions", a two-variable adjunction

$\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$3

is promoted to a Quillen adjunction of two variables by equivalent pushout-product and pullback-product conditions (Mazel-Gee, 2015). The pushout product of maps is

$\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$4

and the Quillen condition requires

$\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$5

together with the corresponding triviality conditions when one factor is in $\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$6 (Mazel-Gee, 2015).

This two-variable formalism is the exact mechanism by which tensor products become homotopically meaningful. A monoidal model $\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$7-category is defined there as a closed monoidal $\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$8-category with a model structure such that its tensor-hom adjunction is a two-variable Quillen adjunction and such that a unit axiom holds via a cofibrant replacement

$\begin{tikzcd} L:(C,\otimes,\mathbb I)\ar[shift left=2]{r} & (D,\wedge,\mathbb J):R \ar[shift left=2]{l}[swap]{\perp}. \end{tikzcd}$9

whose tensor with cofibrant objects induces weak equivalences (Mazel-Gee, 2015). The unit need not itself be cofibrant; the replacement $L$0 acts as a quasi-unit after localization (Mazel-Gee, 2015). This is one of the main places where a weak form of unitality enters the monoidal story.

A further algebraic refinement appears in "Monoidal algebraic model structures", which studies algebraic Quillen two-variable adjunctions and monoidal algebraic model structures (Riehl, 2011). There the pushout-product axiom is enhanced from a closure property of classes of maps to specified coalgebra and algebra structures for algebraic weak factorization systems. This is stronger than weak monoidal Quillen data, but it clarifies what extra categorical structure lies behind ordinary weak monoidal conditions (Riehl, 2011). In particular, the unit condition is expressed via the cofibrant replacement counit

$L$1

and the requirement that

$L$2

be weak equivalences for cofibrant $L$3 (Riehl, 2011).

4. Passage to localizations and symmetric monoidal $L$4-categories

A principal theorem about weak monoidal Quillen equivalences is that they become genuine monoidal equivalences after localization. "Coalgebras in the Dwyer-Kan localization of a model category" proves that if

$L$5

is a weak monoidal Quillen pair between symmetric monoidal model categories with cofibrant units, then the derived functor of $L$6 induces a symmetric monoidal functor

$L$7

If the adjunction is a weak monoidal Quillen equivalence, then $L$8 is a symmetric monoidal equivalence of $L$9-categories (Péroux, 2020).

The proof strategy is explicitly operadic. One passes to operator categories

$X,Y\in C$0

applies symmetric monoidal Dwyer–Kan localization, and then shows that the functor induced by $X,Y\in C$1 preserves coCartesian lifts only after localization, using precisely the weak comparison maps

$X,Y\in C$2

and the unit comparison from a cofibrant replacement of $X,Y\in C$3 (Péroux, 2020). This makes explicit the logic of the weak monoidal condition: comparison morphisms that are not isomorphisms at the model level become equivalences in the localized monoidal $X,Y\in C$4-category.

The $X,Y\in C$5-categorical descent statements of (Mazel-Gee, 2015) fit this theorem as a more general structural backdrop. There one proves that an ordinary Quillen adjunction

$X,Y\in C$6

descends to a derived adjunction

$X,Y\in C$7

that two-variable Quillen adjunctions descend to derived two-variable adjunctions, and that monoidal model $X,Y\in C$8-categories yield canonical closed monoidal localizations (Mazel-Gee, 2015). The paper does not establish a theorem saying that a monoidal left Quillen functor induces a monoidal left derived functor between different monoidal model $X,Y\in C$9-categories, but it does prove the descent of the tensor product itself (Mazel-Gee, 2015). This suggests that weak monoidal Quillen pairs should be seen as a model-level sufficient condition for a derived monoidal functor whose genuine coherence only appears after localization.

5. Operads, monoids, and algebraic structures

Weak monoidal Quillen pairs are especially important for transferring structured algebraic objects. "Equivalence of Operads over Symmetric Monoidal Categories" shows that if

$L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$0

is a weak monoidal Quillen pair and a Quillen equivalence, then the induced adjunction on connected operads

$L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$1

is a Quillen equivalence between semi-model categories (Nguemo, 2018). The right adjoint $L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$2 is defined objectwise on symmetric sequences, and weak monoidality is first lifted from the base categories to symmetric sequences, then used to compare the naive objectwise left adjoint $L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$3 with the actual operadic left adjoint $L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$4 (Nguemo, 2018).

The decisive point is that free operads are built from iterated tensor products indexed by trees, so strict monoidality of the left adjoint is unnecessary provided the comparison maps are weak equivalences on cofibrant inputs (Nguemo, 2018). This is an archetypal application of the concept: weak monoidal Quillen data is exactly enough to transport operadic homotopy theory even when the composition product is technically more complicated than an ordinary symmetric monoidal tensor.

A different but related perspective comes from "Smith Ideals of Operadic Algebras in Monoidal Model Categories" (White et al., 2017). That paper does not use the weak monoidal terminology, but studies a cokernel–kernel adjunction

$L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$5

in which the left adjoint is strong symmetric monoidal and the right adjoint is lax symmetric monoidal, then lifts this adjunction to operadic algebra categories and proves Quillen equivalences under stability and admissibility hypotheses (White et al., 2017). It therefore provides examples where stronger monoidal Quillen data leads to the same type of algebraic transport that weak monoidal Quillen pairs are designed to permit.

At the $L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$6-categorical level, (Péroux, 2020) further shows that weak monoidal Quillen equivalences induce equivalences of categories of $L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$7-algebras and $L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$8-coalgebras: $L(X\otimes Y)\longrightarrow L(X)\wedge L(Y)$9 and likewise for coalgebras (Péroux, 2020). This is especially significant because strict coalgebra model categories often fail to rigidify homotopy coherent coalgebras even when the underlying symmetric monoidal $D$0-categories are equivalent (Péroux, 2020).

6. Examples, limitations, and adjacent frameworks

The paper (Péroux, 2020) treats the Dold–Kan correspondence as a basic example. Depending on which adjunction is chosen, the normalization functor $D$1 may appear as a right adjoint that is lax symmetric monoidal via the shuffle map, or as a left adjoint that is lax comonoidal via the Alexander–Whitney map, though not symmetric (Péroux, 2020). The point is that both adjunctions satisfy the weak monoidal Quillen equivalence conditions with cofibrant units, so the derived Dold–Kan correspondence becomes an equivalence of symmetric monoidal $D$2-categories (Péroux, 2020). This makes Dold–Kan the canonical example of a multiplicative homotopy theory that is only weakly monoidal at the model level but strictly monoidal after localization.

A closely related categorical precursor is Shoikhet’s "A bialgebra axiom and the Dold-Kan correspondence" (Shoikhet, 2011). That paper is not about Quillen pairs, but introduces a bialgebra axiom for a functor carrying both a colax-monoidal structure

$D$3

and a lax-monoidal structure

$D$4

and proves that the Alexander–Whitney and shuffle maps on normalized chains satisfy this axiom strictly (Shoikhet, 2011). The paper explicitly situates itself as arising from attempts to understand weak monoidal Quillen pairs, and it isolates the categorical compatibility needed for passage to monoids and adjoints (Shoikhet, 2011). This suggests that weak monoidal Quillen theory has a purely monoidal-categorical component independent of model structures.

Several papers clarify the limitations of the concept. "Model $D$5-categories II: Quillen adjunctions" does not define weak monoidal Quillen pairs and proves no theorem asserting that a lax or strong monoidal left Quillen functor yields a monoidal derived functor between different monoidal model $D$6-categories (Mazel-Gee, 2015). "Symmetric monoidal categories and $D$7-categories" proves a Quillen equivalence between coherently commutative monoidal $D$8-categories and permutative categories, but does not establish the needed comparison maps to call the adjunction a weak monoidal Quillen pair, and moreover notes that $D$9 is not a symmetric monoidal closed model category (Sharma, 2018). "Monoidal Properties of Franke's Exotic Equivalence" proves monoidal compatibility

$\lambda:c\mathbb I \xrightarrow{\sim} \mathbb I$0

at the level of homotopy categories for an exotic equivalence not induced by any Quillen functor (Nikandros et al., 2023). A plausible implication is that the derived monoidal conclusion of weak monoidal Quillen theory can occur even without an underlying Quillen pair, but then the model-categorical comparison maps are absent.

Finally, weak monoidal ideas also appear in generalized settings. "Weak model categories in classical and constructive mathematics" develops Quillen pairs for weak model categories and a substantial theory of monoidal weak model structures, but does not define a weak monoidal Quillen pair explicitly (Henry, 2018). "Almost mathematics of pointed symmetric monoidal model categories by Smith ideal theory" uses the term lax Quillen monoidal in the Schwede–Shipley sense and proves that certain Quillen equivalences built from homotopically idempotent Smith ideals satisfy these weak monoidal conditions (Kato, 2023). These examples indicate that the concept extends naturally beyond ordinary cofibrantly generated model categories, though the exact terminology varies.

Weak monoidal Quillen pairs thus mark a precise threshold of multiplicative compatibility: strong enough to preserve homotopical monoidal structure after localization, weak enough to apply in settings where strict monoidality fails. Their principal significance lies not in strict tensor preservation at the model level, but in ensuring that the derived functor is monoidal where homotopy theory actually lives (Péroux, 2020).