15
$\begingroup$

Let $X\rightleftarrows Y\circlearrowleft$ be a strong deformation retraction of chain complexes (a.k.a. contraction), i.e. $X\rightarrow Y\rightarrow X$ is the identity, $Y\rightarrow Y$ is a homotopy between the identity and $Y\rightarrow X\rightarrow Y$, and a couple of less relevant vanishing formulas hold. Moreover, let $\mathcal O$ be a Koszul DG-operad and $\mathcal O_\infty=\Omega\mathcal O^{¡}$ its resolution. It is well known how to transfer an $\mathcal O_\infty$ algebra structure on the big complex $Y$ to the small complex $X$, even with explicit formulas in terms of trees due to Kontsevich and Soibelman. Do you know of any written account where this is done in the opposite direction, that is, transferring an $\mathcal O_\infty$ algebra structure on the small complex $X$ to the big complex $Y$?

I'd like to remark that I'm interested in Kontsevich-Soibelman-like formulas. Otherwise, the existence of such transfers follows easily from standard model category arguments. Please, assume that we're working over a field of characteristic zero, if this simplifies things.

$\endgroup$

1 Answer 1

3
$\begingroup$

Let us denote by $p\colon Y\to X$ and $i\colon X\to Y$ the maps of your SDR. Since $pi=\mathop{\mathrm{id}}\nolimits_X$, the map $i$ is injective, and is an isomorphism with its image. The map $\pi=i\circ p$ is a projector (that is, $\pi^2=\pi$); this projector implements a splitting $Y=\ker(\pi)\oplus\mathop{\mathrm{Im}}(\pi)$. Of course, the image of $\pi=i\circ p$ coincides with the image of $i$ since $\pi\circ i=i\circ\pi\circ i=i$.

Recall that for a Koszul operad $\mathcal{O}$, the operad $\mathcal{O}_\infty$ is the cobar construction of the Koszul dual cooperad $\mathcal{O}^{\ ¡}$, and the structure of a $\mathcal{O}_\infty$-algebra on a chain complex $V$ is equivalent to a Maurer-Cartan element in the dg Lie algebra $$\mathfrak{g}_{\mathcal{O},V}:=\prod_{n\ge 1}\mathop{\mathrm{Hom}}\nolimits_{\mathbb{S}}(\overline{\mathcal{O}^{\ ¡}}(n),\mathop{\mathrm{End}}\nolimits_V(n)).$$ Let $\alpha\in\mathfrak{g}_{\mathcal{O},X}$ be the Maurer-Cartan element encoding the given $\mathcal{O}_\infty$-algebra structure on $X$. Consider the map of dg Lie algebras $$ i_*\colon \mathfrak{g}_{\mathcal{O},X}\to \mathfrak{g}_{\mathcal{O},i(X)}=\mathfrak{g}_{\mathcal{O},\pi(Y)}\subset \mathfrak{g}_{\mathcal{O},Y}. $$ This map sends a Maurer-Cartan element $\alpha$ to some Maurer-Cartan element in the algebra $\mathfrak{g}_{\mathcal{O},Y}$, which corresponds to an $\mathcal{O}_\infty$-algebra structure on $Y$. (This is almost verbatim the construction used when dealing with minimal models for $\mathcal{O}_\infty$-algebras. The difference between the two is that if $X$ is not minimal, then $\ker(\pi)$ might be just a part of the space on which the structure would vanish.)

$\endgroup$
10
  • $\begingroup$ Hi Vladimir, I can't find any 10.3.12. I've tried to figure out what result you mean in that chapter without much success, maybe theorem 10.3.7? $\endgroup$ Commented Aug 13, 2015 at 13:23
  • $\begingroup$ Hi Fernando, I refer to Theorem 10.3.12 (p.310) in the electronic version. (math.unice.fr/~brunov/Operads.pdf) $\endgroup$ Commented Aug 13, 2015 at 15:21
  • 2
    $\begingroup$ Ok, that's theorem 10.3.7 in the published version. The proof is mistaken, if the argument were true any quasi-isomorphism should be injective! Actually, what I'm asking about would yield a proof of that theorem as a corollary (at least over a field), since any quasi-isomorphism between cofibrant complexes can be replaced with a roof of strong deformation retractions. $\endgroup$ Commented Aug 13, 2015 at 15:57
  • 1
    $\begingroup$ @FernandoMuro Is it actually much simpler than we (I) think? In your picture the map $i\colon X\to Y$ must be injective and is an isomorphism with its image. Let us now use the projector $\pi=i\circ p$ to split $Y$ as the direct sum of its kernel and its image. (The latter coincides with the image of $i$. The former, we do not care.) Then let us argue like in the minimal model theorem: if we transfer the structure that we have on $X$ to the image of $\pi$ via the isomorphism $i$, and let all yet undefined structure maps be zero, we get an O-infinity structure on $Y$. Or something fails? $\endgroup$ Commented Aug 13, 2015 at 16:36
  • 1
    $\begingroup$ $\prod$ instead of $\bigoplus$? $\endgroup$ Commented Aug 14, 2015 at 12:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .