This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Non-unital algebra objects of stable symmetric monoidal model categories by Smith ideal theory

Yuki Kato National institute of technology, Ube college, 2-14-1, Tokiwadai, Ube, Yamaguchi, JAPAN 755-8555. [email protected]
Abstract.

This note remarks that the correspondence between non-unital algebras and augmented unital algebras can be derived from Hovey’s Smith ideal theory. Applying Smith ideal theory of stable symmetric monoidal model category, we formulate non-unital algebra objects of stable symmetric monoidal model categories and generalize the correspondence between non-unital algebra objects and augmented algebra objects.

Key words and phrases:
Symmetric monoidal model categories, Smith ideals, Non-unital algebras

1. Introduction

A ring without the assumption existence of a multiplicative unit 11 is called a non-unital ring. Any unital ring is regarded as a non-unital ring by forgetting the existence of 11. This paper fixes a base unital ring VV and let AlgVnu\mathrm{Alg}^{\rm nu}_{V} denote the category of non-unital VV-algebras. An augmentation εA:AV\varepsilon_{A}:A\to V is a ring homomorphism that the unit homomorphism VAV\to A is a section of εA\varepsilon_{A}, and a VV-algebra with an augmentation is called an augmented VV-algebra, denoting AlgV//V\mathrm{Alg}_{V//V} the category of augmented VV-algebras. It is well-known that there exists a categorical equivalence between AlgVnu\mathrm{Alg}^{\rm nu}_{V} and AlgV//V\mathrm{Alg}_{V//V}: For any VV-algebra AA, the direct sum VAV\oplus A has a canonical unital ring structure defined by

(m,a)(n,b)=(mn,na+mb+ab)(m,\,a)\cdot(n,\,b)=(mn,\,na+mb+ab)

for m,nVm,\,n\in V and a,bAa,\,b\in A. It is easily checked that the functor the adjunction

V():AlgVnuAlgV//V:Ker(ϵ():V)V\oplus(-):\mathrm{Alg}^{\rm nu}_{V}\rightleftarrows\mathrm{Alg}_{V//V}:\mathrm{Ker}(\epsilon_{(-)}:-\to V)

is a pair of categorical equivalences, where Ker(ε():V)\mathrm{Ker}(\varepsilon_{(-)}:-\to V) is the kernel functor of augmentations.

The main result of this paper is a model categorical analogue of the above categorical equivalence (Theorem 3.4): A stable model category is a model category whose homotopy category is triangulated. We generalize the definition of non-unital algebras for stable symmetric monoidal model categories by using Hovey’s Smith ideal theory [Hov14, Theorem 4.3]. An advantage of Smith ideal theory of stable symmetric monoidal model categories is that it enables us to clarify that a symmetric monoidal model structure of non-unital algebra objects is the push-out products induced by ones.

Acknowledgements

The author was supported by Grants-in-Aid for Scientific Research No.23K03080, Japan Society for the Promotion of Science.

2. Smith ideal theory of symmetric monoidal model categories

We explain Hovey [Hov14]’s Smith ideal theory of symmetric monoidal categories. This section assumes that categories are always pointed, and 0 denotes the zero object.

2.1. Smith ideal theory of pointed categories

For any symmetric monoidal category 𝒞\mathcal{C}, let Alg(𝒞)\mathrm{Alg}(\mathcal{C}) denote the subcategory of 𝒞\mathcal{C} spanned by monoid objects in the sense of MacLane [Mac88].

Let [1][1] denote the category with two objects 0 and 11, and only one non-identity morphism 010\to 1. For any pointed category 𝒞\mathcal{C}, the diagram category Fun([1],𝒞)\mathrm{Fun}([1],\,\mathcal{C}) is called the arrow category, and Ar(𝒞)\mathrm{Ar}(\mathcal{C}) denotes it. The symmetric monoidal structure of 𝒞\mathcal{C} inherits Ar(𝒞)\mathrm{Ar}(\mathcal{C}) two symmetric monoidal structures: For any two morphisms in 𝒞\mathcal{C}, f:X0X1f:X_{0}\to X_{1} and g:Y0Y1g:Y_{0}\to Y_{1}, the tensor product monoidal structure is defined by fgf\otimes g as the composition

fg:X0Y0X1Y1,f\otimes g:X_{0}\otimes Y_{0}\to X_{1}\otimes Y_{1},

and another is the push-out product monoidal structure defined by the induced morphism:

fg:(X0Y1)X0Y0(X1Y0)X1Y1.f\Box g:(X_{0}\otimes Y_{1})\amalg_{X_{0}\otimes Y_{0}}(X_{1}\otimes Y_{0})\to X_{1}\otimes Y_{1}.

We let Ar(𝒞)\mathrm{Ar}^{\otimes}(\mathcal{C}) the arrow category with the tensor product monoidal structure and Ar(𝒞)\mathrm{Ar}^{\Box}(\mathcal{C}) with the push-out monoidal structure. Hovey proved the cokernel functor is strongly monoidal and admits a lax monoidal right adjoint:

Theorem 2.1 ([Hov14] Theorem 1.4).

Let 𝒞\mathcal{C} be a pointed closed symmetric monoidal category. Then the functor

cok:Ar(𝒞)(f:XY)(cok(f):YCoker(f))Ar(𝒞)\mathrm{cok}:\mathrm{Ar}^{\Box}(\mathcal{C})\ni(f:X\to Y)\mapsto(\mathrm{cok}(f):Y\to\mathrm{Coker}(f))\in\mathrm{Ar}^{\otimes}(\mathcal{C})

is strongly symmetric monoidal, and its right adjoint is the kernel functor ker:(f:XY)(ker(f):Ker(f)X)\mathrm{ker}:(f:X\to Y)\mapsto(\mathrm{ker}(f):\mathrm{Ker}(f)\to X). \Box

Hovey mentioned in [Hov14], in general, the kernel functor ker:Ar(𝒞)Ar(𝒞)\mathrm{ker}:\mathrm{Ar}(\mathcal{C})\to\mathrm{Ar}(\mathcal{C}) is lax symmetric monoidal: A canonical functor

ker(f)Ker(g)ker(fg)\mathrm{ker}(f)\Box\mathrm{Ker}(g)\to\mathrm{ker}(f\otimes g)

is induced by the canonical morphism cok(ker(f)Ker(g))cok(ker(f))cok(ker(g))fg\mathrm{cok}(\mathrm{ker}(f)\Box\mathrm{Ker}(g))\simeq\mathrm{cok}(\mathrm{ker}(f))\otimes\mathrm{cok}(\mathrm{ker}(g))\to f\otimes g for any f,gAr(𝒞)f,\,g\in\mathrm{Ar}^{\otimes}(\mathcal{C}). Hence those functors cok\mathrm{cok} and ker\mathrm{ker} send monoid objects to monoid objects. A Smith ideal is a monoid object of the symmetric monoidal category Ar(𝒞)\mathrm{Ar}(\mathcal{C}) with respect to the push-out product monoidal structure.

2.2. Definition of non-unital algebra objects of symmetric monoidal categories

For any pointed closed symmetric monoidal category 𝒞\mathcal{C} with a monoidal unit VV, let Alg(𝒞)V//V\mathrm{Alg}(\mathcal{C})_{V//V} denote the full subcategory of Ar(𝒞)\mathrm{Ar}^{\otimes}(\mathcal{C}) spanned by augmentations of algebra objects of the tensor product monoidal structure.

Let 𝒞\mathcal{C} be a pointed closed symmetric monoidal category and Arim(𝒞)\mathrm{Ar}^{\rm im}(\mathcal{C}) denotes the full subcategory spanned those objects f:RSf:R\to S such that the unit fker(cok(f))f\to\mathrm{ker}(\mathrm{cok}(f)) is an isomorphism. Then the cokernel functor Arim(𝒞)Arcoim(𝒞)\mathrm{Ar}^{\rm im}(\mathcal{C})\to\mathrm{Ar}^{\rm coim}(\mathcal{C}) is a categorical equivalence.

Proposition 2.2.

Let 𝒞\mathcal{C} be a locally presentable abelian symmetric monoidal category. Then the arrow category Ar(𝒞)\mathrm{Ar}(\mathcal{C}) is also locally presentable and there exist reflective localization functors Lim:Ar(𝒞)Arim(𝒞)L_{\rm im}:\mathrm{Ar}^{\Box}(\mathcal{C})\to\mathrm{Ar}^{\rm im}(\mathcal{C}) by the unit idim\mathrm{id}\to\mathrm{im} and Lcoim:Ar()Arcoim()L_{\rm coim}:\mathrm{Ar}^{\otimes}(\mathcal{M})\to\mathrm{Ar}^{\rm coim}(\mathcal{M}) by the counit coimid\mathrm{coim}\to\mathrm{id} such that the adjunction

cok:Ar(𝒞)Ar(𝒞):ker\mathrm{cok}:\mathrm{Ar}^{\Box}(\mathcal{C})\rightleftarrows\mathrm{Ar}^{\otimes}(\mathcal{C}):\mathrm{ker}

induces categorical equivalences

cok:Arim(𝒞)Arcoim(𝒞):ker.\mathrm{cok}:\mathrm{Ar}^{\rm im}(\mathcal{C})\rightleftarrows\mathrm{Ar}^{\rm coim}(\mathcal{C}):\mathrm{ker}.
  • proof.

    Since any abelian category is binormal, fim(f)f\to\mathrm{im}(f) is an isomorphism if and only if ff is a monomorphism. By Adámek–Rosický [AR94, p.44, Corollary 1.5.4], the arrow category is locally presentable, and it admits reflective localization. By the definitions of those functors: im=kercok\mathrm{im}=\mathrm{ker}\circ\mathrm{cok} and coim=cokker\mathrm{coim}=\mathrm{cok}\circ\mathrm{ker}, the restriction cok:Arim(𝒞)Arcoim(𝒞)\mathrm{cok}:\mathrm{Ar}^{\rm im}(\mathcal{C})\to\mathrm{Ar}^{\rm coim}(\mathcal{C}) are quasi-inverse functors. \Box

Definition 2.3.

Let 𝒞\mathcal{C} be a locally presentable symmetric monoidal abelian category with a monoidal unit VV. A non-unital monoid object of 𝒞\mathcal{C} is a Smith ideal j:IRj:I\to R in the localized full subcategory Arim(𝒞)\mathrm{Ar}^{\rm im}(\mathcal{C}) satisfying that the cokernel cok(j):RCoker(j)\mathrm{cok}(j):R\to\mathrm{Coker}(j) is isomorphic with an argumentation εR:RV\varepsilon_{R}:R\to V, and Algnu(𝒞)\mathrm{Alg}^{\rm nu}(\mathcal{C}) denote the full subcategory of Alg(Ar(𝒞))\mathrm{Alg}(\mathrm{Ar}^{\Box}(\mathcal{C})) spanned by non-unital algebra objects, where Ar(𝒞)\mathrm{Ar}^{\Box}(\mathcal{C}) denotes the arrow category of 𝒞\mathcal{C} whose monoidal structure is the push-out monoidal structure.

Proposition 2.4.

Further, let Algnu(𝒞)\mathrm{Alg}^{\rm nu}(\mathcal{C}) denote the full subcategory of Ar(𝒞)\mathrm{Ar}^{\Box}(\mathcal{C}) the full subcategory which is the essential image of the restriction of the kernel functor on Alg(𝒞)V//V\mathrm{Alg}(\mathcal{C})_{V//V}. Then any object j:IRj:I\to R of Algnu(𝒞)\mathrm{Alg}^{\rm nu}(\mathcal{C}), the unit jim(j)j\to\mathrm{im}(j) is an isomorphism.

  • proof.

    By definition of the category Algnu(𝒞)\mathrm{Alg}^{\rm nu}(\mathcal{C}), one has an isomorphism VIRV\oplus I\to R. The assertion is clear. \Box

Corollary 2.5.

Let 𝒞\mathcal{C} be a locally presentable symmetric monoidal abelian model category. The adjunction

cok:Ar(𝒞)Ar(𝒞):ker\mathrm{cok}:\mathrm{Ar}^{\Box}(\mathcal{C})\rightleftarrows\mathrm{Ar}^{\otimes}(\mathcal{C}):\mathrm{ker}

induces categorical equivalences

cok:Algnu(𝒞)Alg(𝒞)V//V:ker.\mathrm{cok}:\mathrm{Alg}^{\rm nu}(\mathcal{C})\to\mathrm{Alg}(\mathcal{C})_{V//V}:\mathrm{ker}.

\Box

3. Non-unital commutative algebra objects of symmetric monoidal model categories

A symmetric monoidal model category \mathcal{M} is a model category with a symmetric monoidal structure :×-\otimes-:\mathcal{M}\times\mathcal{M}\to\mathcal{M} such that, for any object MM of \mathcal{M}, those functors ()M(-)\otimes M and M()M\otimes(-) are left Quillen functors on \mathcal{M}.

3.1. The arrow categories of pointed symmetric monoidal model categories

The category Ar()\mathrm{Ar}(\mathcal{M}) has two canonical model structures, the injective model structure and the projective model structure induced by \mathcal{M}’s:

Definition 3.1.

Let \mathcal{M} be a model category. The arrow category Ar()\mathrm{Ar}(\mathcal{M}) has the following two model structures.

  • (Injective model structure) A morphism α:(f:X0X1)(g:Y0Y1)\alpha:(f:X_{0}\to X_{1})\to(g:Y_{0}\to Y_{1}) is a cofibrations (resp. weak equivalence) in Ar()\mathrm{Ar}(\mathcal{M}) if and only if so is each Evi(α)\mathrm{Ev}_{i}(\alpha) for i=0, 1i=0,\,1. Fibrations are morphisms with the right lifting property for all trivial cofibrations.

  • (Projective model structure) A morphism α:(f:X0X1)(g:Y0Y1)\alpha:(f:X_{0}\to X_{1})\to(g:Y_{0}\to Y_{1}) is a fibrations (resp. weak equivalence) in Ar()\mathrm{Ar}(\mathcal{M}) if and only if so is each Evi(α)\mathrm{Ev}_{i}(\alpha) for i=0, 1i=0,\,1. Cofibrations are morphisms with the right lifting property for all trivial fibrations.

In a pointed model category \mathcal{M}, we consider a homotopically commutative diagram:

X\textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}f\scriptstyle{f}Y\textstyle{Y\ignorespaces\ignorespaces\ignorespaces\ignorespaces}g\scriptstyle{g}0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Z.\textstyle{Z.}

If the diagram is a homotopy Cartesian square, then XX is said to be a homotopy kernel of gg, and if it is a homotopy coCartesian square, then ZZ is a homotopy cokernel of ff. We can consider homotopy image objects and homotopy coimage objects as additive categories.

On the arrow category Ar()\mathrm{Ar}(\mathcal{M}), those functors cok:Ar()Ar()\mathrm{cok}:\mathrm{Ar}(\mathcal{M})\to\mathrm{Ar}(\mathcal{M}) and ker:Ar()Ar()\mathrm{ker}:\mathrm{Ar}^{\otimes}(\mathcal{M})\to\mathrm{Ar}^{\Box}(\mathcal{M}) are defined as follows: For a morphism f:XYf:X\to Y in \mathcal{M}, the arrow cok(f)\mathrm{cok}(f) is YCokerfY\to\mathrm{Coker}f and ker(f)\mathrm{ker}(f) Ker(f)X\mathrm{Ker}(f)\to X. Then the pair

cok:Ar()Ar():ker\mathrm{cok}:\mathrm{Ar}(\mathcal{M})\rightleftarrows\mathrm{Ar}(\mathcal{M}):\mathrm{ker}

is a Quillen adjunction.

Definition 3.2.

Let \mathcal{M} be a pointed symmetric monoidal model category. A Smith ideal in \mathcal{M} is a monoid object j:IRj:I\to R in the symmetric monoidal model category Ar()\mathrm{Ar}^{\Box}(\mathcal{M}) with respect to the push-out product monoidal model structure.

We say that a Smith ideal j:IRj:I\to R is unit cokernel if the cokernel of jj is isomorphic to the monoidal unit object VV and cok(j):RCoker(j)\mathrm{cok}(j):R\to\mathrm{Coker}(j) is an augmentation of the unit morphism VRV\to R. Let Algnu(𝒞)\mathrm{Alg}^{\rm nu}(\mathcal{C}) denote the full subcategory of Alg(Ar(𝒞))\mathrm{Alg}(\mathrm{Ar}^{\Box}(\mathcal{C})) spanned by unit cokernel Smith ideals.

Definition 3.3.

Let \mathcal{M} be a stable symmetric monoidal model category with a monoidal unit VV and Algnu()\mathrm{Alg}^{\rm nu}(\mathcal{M}) denote the full subcategory Alg(Ar())\mathrm{Alg}(\mathrm{Ar}^{\Box}(\mathcal{M})) spanned by Smith ideals whose cokernels are weakly equivalent to VV. We say that an object of Algnu()\mathrm{Alg}^{\rm nu}(\mathcal{M}) is a non-unital commutative algebra object of \mathcal{M}.

Theorem 3.4.

Let \mathcal{M} be a stable symmetric monoidal model category with a monoidal unit VV. Then the Quillen equivalence

cok:Ar()Ar():ker\mathrm{cok}:\mathrm{Ar}(\mathcal{M})\rightleftarrows\mathrm{Ar}(\mathcal{M}):\mathrm{ker}

induces a left Quillen equivalence between cok:Algnu()Alg()V//V\mathrm{cok}:\mathrm{Alg}^{\rm nu}(\mathcal{M})\to\mathrm{Alg}(\mathcal{M})_{V//V}.

  • proof.

    By using the second statement of [Hov14, Theorem 4.3], the cokernel functor cok:Ar()Ar()\mathrm{cok}:\mathrm{Ar}(\mathcal{M})\to\mathrm{Ar}(\mathcal{M}) induces the left Quillen functor cok:Algnu()Alg()V//V\mathrm{cok}:\mathrm{Alg}^{\rm nu}(\mathcal{M})\to\mathrm{Alg}(\mathcal{M})_{V//V}, being essentially surjective on the homotopy categories by definition. Since the unit u:IdAr()kercoku:\mathrm{Id}_{\mathrm{Ar}(\mathcal{M})}\to\mathrm{ker}\circ\mathrm{cok} is a weak equivalence by [Hov14, Theorem 4.3] again, the induced functor cok:Algnu()Alg()V//V\mathrm{cok}:\mathrm{Alg}^{\rm nu}(\mathcal{M})\to\mathrm{Alg}(\mathcal{M})_{V//V} is homotopically fully faithful. \Box

Remark 3.5.

If \mathcal{M} is a locally presentable stable symmetric monoidal model category, the presentable \infty-category of augmented algebra is represented by the model category Alg()V//V\mathrm{Alg}(\mathcal{M})_{V//V}. Hence the left-hand-side Algnu()\mathrm{Alg}^{\rm nu}(\mathcal{M}) is equivalent to the \infty-category defined by Lurie [Lur17, p.949, Definition 5.4.4.9 and Proposition 5.4.4.10].

References

  • [1]
  • [AR94] Adámek, Jiří ; Rosický, Jiří: Lond. Math. Soc. Lect. Note Ser.. Bd. 189: Locally presentable and accessible categories. Cambridge: Cambridge University Press, 1994. – ISBN 0–521–42261–2
  • [Hir03] Hirschhorn, Philip S.: Math. Surv. Monogr.. Bd. 99: Model categories and their localizations. Providence, RI: American Mathematical Society (AMS), 2003. – ISBN 0–8218–3279–4
  • [Hov99] Hovey, Mark: Mathematical Surveys and Monographs. Bd. 63: Model categories. American Mathematical Society, Providence, RI, 1999. – xii+209 S. – ISBN 0–8218–1359–5
  • [Hov14] Hovey, Mark: Smith ideals of structured ring spectra. Available at:https://arxiv.org/abs/1401.2850, 2014
  • [Lur17] Lurie, Jacob: Higher Algebra. available at:https://www.math.ias.edu/ lurie/papers/HA.pdf, 2017
  • [Mac88] MacLane, Saunders: Grad. Texts Math.. Bd. 5: Categories for the working mathematician. 4th corrected printing. New York etc.: Springer-Verlag, 1988. – ISBN 3–540–90035–7