related by the Dold-Kan correspondence
By definition L-∞ algebras are the ∞-algebras in the category of chain complexes over the Lie operad. As such they carry a model structure on algebras over an operad. There is a strictification which leads equivalently to a model structure on dg-Lie algebras.
A more geometric way is to think of L-∞ algebras as being the tangent spaces to connected smooth ∞-groupoids, hence to the delooping/moduli ∞-stacks of smooth ∞-groups, hence as the first order infinitesimal neighbourhood
of the essentially unique point (see at Lie differentiation). Since this is equivalently the first order neighbourhood of the formal neighbourhood (the jets) and the formal neighbourhood can be described purely in terms of associative algebra/coalgebra (see at smooth algebra) many models for -algebras are formulated in terms of such data.
Conversely, the category of L-∞ algebras (and general “weak” morphisms between them) is the full subcategory of that of counital cocommutative dg-coalgebras on those whose underlying bare graded-commutative coalgebra (forgetting the codifferential) is free
Accordingly it is of interest to have also model structure on dg-coalgebras or dually (on pro-objects of) dg-algebras which presents -algebras. Prop. 9 below identifies the category as the (full sub-)category of fibrant objects inside such a model structure of “differential graded formal spaces”, which in turn is related by a zig-zag of Quillen equivalence to various other models
So we write “” here for the 1-category of -algebras and general morphisms between them, since this is an entry on model category presentations. If we want to refer to the (∞,1)-category of -algebras we here write explicitly “”, referring to the simplicial localization of this 1-category.
All gradings in the following are -gradings, unless explicitly stated otherwise. In terms of the underlying geometry this means that we are dealing with derived geometry (see below the section Simplicial sheaves over comsimplicial formal spaces for details): the algebra elements in positive degree correspond to categorical/simplicial/∞-groupoid/∞-stack-degree, and those in negative degree to the cosimplicial degree of the derived site of cosimplicial formal spaces.
Technically this affects for instance the nature of fibrations: for instance the model structure on dg-Lie algebras below is transferred from a model structure on chain complexes. For unbounded chain complexes this is the “Categorical projective class structure” whose fibrations are the chain maps that are surjective in every degree. This appears for instance in prop. 2 and prop. 10 below.
On the other hand, if one considered chain complexes in non-negative degree (for tangent complexes in “higher but non-derived geometry”), then one would use the Projective structure on chain complexes in non-negative degree. This has as fibrations precisely the chain maps that are surjective in every positive degree. This case is (currently) not discussed in the following.
Some of the model structures below are on the category of -algebras with “strict” morphisms between them, namely for those morphisms which are morphisms of algebras over an operad for an -algebra regarded as an algebra over a cofibrant resolution of the Lie operad. We write
for the wide subcategory on the strict -morphisms.
As their name indicates, L-∞ algebras are the homotopy algebras over the Lie operad (in a category of chain complexes). As such, the general theory of model structures on algebras over an operad provides a model category structure on the category of -algebras.
This we discuss here. But there is also a natural identification of -algebras with infinitesimal derived ∞-stacks. For expressing this a host of other, Quillen equivalent model structures are available. These we discuss below in Definitions as formal/infinitesimal ∞-stacks.
By the general discussion at model structure on dg-algebras over an operad, if is a field which contains the field of rational numbers, then for every symmetric operad (uncolored) in the category of chain complexes (unbounded) , the free-forgetful adjunction
the weak equivalences are the morphisms that are quasi-isomorphisms on the (underlying) chain complexes;
the fibrations are the morphisms that are degreewise surjections on the (underlying) chain complexes.
Hence in particular is Quillen adjunction between these model structures.
So this is in particular true for the standard cofibrant resolution of the Lie operad. In this case is the category of (unbounded) -algebras (with strict -maps between them as in remark 3 above) and hence is equipped with a transferred model structure this way
We list here definitions of various further model category structures that all present the (∞,1)-category of L-∞ algebras and describe a web of zig-zags of Quillen equivalences between them. These Quillen equivalences may be thought of as presenting an equivalence between the (∞,1)-category of -algebras and that of infinitesimal derived ∞-stacks (“formal moduli problems”).
The following tabulates the main categories considered below, the functors relating them and their homotopy theoretic nature. The last row points to the relevant definitions and propositions of the following text.
|L-∞ algebras||form Chevalley-Eilenberg algebra||pro-objects in commutative Artin dg-algebras||dualize||commutative dg-coalgebra||form tangents||dg-Lie algebras|
|category of fibrant objects||equivalence of (∞,1)-categories under simplicial localization||opposite model structure of cofibrantly generated model category||left Quillen equivalence||model category||left Quillen equivalence||cofibrantly generated model category|
|prop. 1||def. 8||def. 7||prop. 14||def. 3||prop. 3||def. 1|
Here we are trying to use suggestive names of the categories involved. The notation used here corresponds to that in (Pridham) by the following dictionary
(handle with care, may still need attention)
|notation used here||notation in Pridham|
|, def. 6||, def. 1.32|
|, def. 7||. def. 3.1|
|, def. 4.6|
We call this the model structure on dg-Lie algebras.
and then extended as a coderivation.
(The Chevalley-Eilenberg dg-coalgebra.)
If one thinks of a dg-coalgebra as presenting a a derived formal space, as discuss below then its image under , prop. 3, may be thought of as its tangent dg-Lie algebra. Therefore is also called the tangent Lie algebra functor.
There exists a model category structure on for which
We call this the model structure on dg-coalgebras.
Beware that the class of weak equivalences in prop. 4 is not that of quasi-isomorphisms on the chain complexes underlying the dg-coalgebras.
But they form a sub-class:
The Chevalley-Eilenberg functor
The pair of adjoint functors
is a weak equivalence that exhibits as a fibrant resolution and moreover, if was already fibrant, hence by prop. 9 below an L-∞ algebra, as a strictification of : because a dg-Lie algebra is an -algebra in which the Lie bracket satisfies its Jacobi identity strictly (not just up to a homotopy measured by the trinary bracket) and in which the “Jacobiator identity” holds strictly, etc.
for the category of infinitesimally thickened points, the full subcategory of the opposite category of Artin algebras (“Weil alebras” in the language of synthetic differential geometry). The category of infinitesimally thickened points.
In (Pridham) this is def. 1.18.
In (Pridham) this is def. 1.32.
This is (Pridham, def. 2.7, theorem 2.14)
Between quasi-smooth objects in the weak equivalences are precisely the morphisms which are weak homotopy equivalences of simplicial sets over each object in .
This is (Pridham, cor. 2.16).
for its category of pro-objects and write
for the opposite category of that.
This is (Pridham, def. 3.1).
While it so happens that every coalgebra and dg-coalgebra is the filtered colimit of its finite-dimensional subalgebras (see at coalgebra – As filtered colimits), this is not in general the case for algebras. But it follows that the linear dual of a general coalgebra is a filtered limit of finite-dimensional algebras, hence a pro-object in finite dimensional algebras. This is the reason for the appearance of pro-objects in prop. 7.
There is a cofibrantly generated model category structure on , def. 7 – hence an opposite model structure on – whose weak equivalences are those morphisms that are local morphisms relative to quasi-smooth maps in the homotopy category of the slice category over their codomain.
This is (Pridham, prop. 4.36).
This is proven inside the proof of (Pridham, prop. 4.42).
Prop. 9 shows in particular that the category of prop/def. 1 carries the structure of a category of fibrant objects that presents the homotopy theory of -algebras. Notice that, of course, passing to the full subcategory of fibrant objects does not change the homotopy theory presented by the underlying category with weak equivalences in that we have an equivalence of (∞,1)-categories between the simplicial localizations
The following proposition characterizes the structure of this category of fibrant objects.
Beware, as in remark 5, that the class of weak equivalences in prop. 10 differs from that of those maps on associated Chevalley-Eilenberg algebras which are quasi-isos on the underlying chain complexes of the dg-algebra (which instead are the weak equivalences in the standard model structure on dg-algebras, hence in particular those used in Sullivan rational homotopy theory). Instead the weak equivalences correspond to the maps of CE-algebra that are quasi-isomorphisms only on the chain complexes given by the co-unary component of the differential of the CE-algebra.
This is (Pridham, cor. 4.49).
For arbitrary , there is a Quillen equivalence
given by sending an object in , hence an dg-algebra , to its dual dg-coalgebra , is the left adjoint part of a Quillen equivalence between the model structure on , prop. 8, and the model structure on dg-coalgebras, prop. 4.
This equivalence has the nice property that starting with the Chevalley-Eilenberg algebra and then “denormalizing” it under dual monoidal Dold-Kan to a cosimplicial nilpotent algebra yields manifestly an incarnation of the -algebra in terms of simplicial complexes of infinitesimal simplices as is implicit in the work of Anders Kock in synthetic differential geometry. This is spelled out further in dcct, section 4.5.1.
This is (Pridham, def. 4.11, prop. 4.12).
This is (Pridham, theorem 4.26).
We discuss some further properties of the above model category structures.
This observation has been communicated privately by Jonathan Pridham
We need to show that the pullback of a weak equivalence along a fibration is again a weak equivalence. If is a fibration, this is automatic, so by factorisation we reduce to the case where is a cofibration. Now, every trivial cofibration is of a composition of acyclic small extensions, so we may take to be of an acyclic small extension with kernel . Then is of a quasi-free map , so the pullback is of , and , so is also an acyclic small extension.
In any model category we have a notion of homotopy between 1-morphisms. In any category of fibrant objects we still have a notion of right homotopy, given by maps into a path space object. So all of the above model category/fibrant object category structures yield models for homotopies between morphisms of -algebras.
More generally, a description of the full derived hom space between two -algebras is obtained via remark 8 from the description of derived hom-spaces in categories of fibrant objects.
Recognizing homotopy fiber products in any of the model structure above can be a bit subtle. A recognition principle of homotopy fibers over abelian -algebras , hence useful for discussion of ∞-Lie algebra extensions), is described in (Fiorenza-Rogers-Schreiber 13, theorem 3.1.13).
Precursors for 2-reduced dg-algebras are dicussed in
The homotopy-theoretic nature of -algebras and their relation to deformation problems was then notably amplified in
Model structurs on algebras over operads in chain complexes were discussed generally in
The full model structure on dg-coalgebras (in characteristic 0) and the Quillen equivalence of dg-Lie algebras as well as the interpretation in terms of formal -stacks is due to
the relation to -stacks is discussed more in detail.
More model category theoretic developments relating various of the previous approaches and generalizing to arbitrary characteristic are in
A useful summary of that paper is given in the notes, by Stefano Maggiolo.
A discussion of path space objects for -algebras is in section 5 of