7

There is a 1-1-correspondence between $A_\infty$-algebra structures on a graded vector space $A$ and dg. coalgebra structures on the bar construction $(\bar T(sA),\Delta)$.

My question: Is there any analogous statement for $C_\infty$-algebras? Recently I heard that a $C_\infty$-structure on $C$ corresponds to a dg. structure on the cofree Lie coalgebra generetad by $sC$ but I can't find any reference for that or prove it myself.

Najib Idrissi
  • 56,269
  • Yes. The keyword you want to look up is "Koszul duality," more specifically Koszul duality between the Lie operad and the commutative operad (whereas the first statement is Koszul duality between the associative operad and itself). I don't know a reference, but maybe look up references about operads. – Qiaochu Yuan Feb 27 '16 at 16:48
  • Qiaochu Yuan: Thanks for answer! Since the first correspondence can be explained without knowing anything about Koszul duality, I hoped that the second correspondence could be explained in a similar manner. – Bashar Saleh Feb 27 '16 at 18:30

1 Answers1

7

As Qiaochu Yuan mentions in a comment, Koszul duality of operads is an answer. I think the standard reference now is Algebraic operads by Loday and Vallette. It's a rather vast subject, so I'll try to make a (quick) sketch of it – read the book for more details. There is also the book Operads in algebra, topology and physics by Markl, Shnider, and Stasheff.

For all this answer, I will consider things happening in dg-modules (chain complexes) over a field.

Koszul duality is something that takes a (quadratic) operad $\mathtt{P}$ and spits out a dual cooperad $\mathtt{P}^¡$ together with a so-called twisting morphism $\kappa : \mathtt{P}^¡ \to \mathtt{P}$. This twisting morphism induces a morphism from the cobar construction $\Omega \mathtt{P}^¡$ (an operad) to $\mathtt{P}$.

When the operad $\mathtt{P}$ satisfies a special property called "being Koszul", this morphism $\Omega \mathtt{P}^¡ \to \mathtt{P}$ is a quasi-isomorphism, i.e. it induces an isomorphism on homology. The operad $\Omega \mathtt{P}^¡$ is then often denoted $\mathtt{P}_\infty$.

It's sometimes a bit easier to work with operads rather than with cooperads, and it's possible to construct a Koszul dual operad $\mathtt{P}^!$ from $\mathtt{P}$. This is an involution: $(\mathtt{P}^!)^! = \mathtt{P}$. The famous trinity of operads $\mathtt{Ass}$ (associative algebras), $\mathtt{Com}$ (commutative algebras), and $\mathtt{Lie}$ (Lie algebras) are all Koszul, and their duals are $\mathtt{Ass}^! = \mathtt{Ass}$, $\mathtt{Com}^! = \mathtt{Lie}$, and $\mathtt{Lie}^! = \mathtt{Com}$.

Now, what's the link with $A_\infty$- and $C_\infty$-algebras? It turns out that an $A_\infty$-algebra is the same thing as an algebra over $\mathtt{Ass}_\infty = \Omega \mathtt{Ass}^{¡}$, and a $C_\infty$-algebra is the same thing as an algebra over $\mathtt{Com}_\infty = \Omega \mathtt{Com}^¡$. This is not really a coincidence: when $\mathtt{P}$ is a Koszul operad, the operad $\mathtt{P}_\infty$ enjoys very nice properties. If $A$ is a dg-module equipped with a $\mathtt{P}$-algebra structure and $B$ is a dg-module quasi-isomorphic to $A$, then $B$ cannot necessarily be equipped with a $\mathtt{P}$-algebra structure; but it can be equipped with a $\mathtt{P}_\infty$, such that the quasi-isomorphism respect this structure. Moreover, a quasi-isomorphism $X \xrightarrow{\sim} Y$ of $\mathtt{P}$-algebras can always be inverted $X \xleftarrow{\sim} Y$, but the inverse is really an $\infty$-quasi-isomorphism of $\mathtt{P}_\infty$-algebras. (I believe these properties are what initially motivated the definition of $A_\infty$- and $C_\infty$-algebas, even before Koszul duality of operads was discovered.)

And now, Koszul duality allows one to reformulate the definition of $A_\infty$- and $C_\infty$-algebras. Since $\mathtt{Ass}^! = \mathtt{Ass}$, by some general abstract nonsense, to give $X$ an algebra structure over $A_\infty = \mathtt{Ass}_\infty$ is exactly the same thing as giving a square zero coderivation on $T^c(\Sigma X)$, the cofree coassociative (conilpotent) coalgebra on the suspension of $X$. And similarly, since $\mathtt{Com}^! = \mathtt{Lie}$, to give $X$ an algebra structure over $C_\infty = \mathtt{Com}_\infty$ is exactly the same thing as giving a square zero coderivation on $L^c(\Sigma X)$, the cofree Lie coalgebra on the suspension of $X$.

(And as a bonus, since $\mathtt{Lie}^! = \mathtt{Com}$, an $L_\infty$ structure on $X$ is thus the same thing as a square zero coderivation on $S^c(\Sigma X)$, the cofree cocommutative coalgebra on the suspension of $X$.)

Najib Idrissi
  • 56,269
  • Maybe it's possible to give an ad-hoc explanation for $C_\infty$ that doesn't mention Koszul duality, but everything falls into place so perfectly when viewed through it... – Najib Idrissi Feb 27 '16 at 21:02
  • Najib Idrissi: Thanks for the answer. I have to digest this a bit – Bashar Saleh Feb 27 '16 at 21:11