(Twisted) canonical supermultiplets and their resolutions as open-closed homotopy algebras

Throughout we work over $\mathbbm{C}$, the field of complex numbers. Tensor products are taken over $\mathbbm{C}$ unless otherwise specified. Differential graded (dg) vector spaces and their morphisms are $\mathbbm{Z}$-graded. The dual of a graded vector space is taken to be the direct sum of the dual of each graded piece.

Throughout we work with the Koszul sign convention. We let $\tau_{V,W}:V\otimes W\to W\otimes V$ denote the flip map. Similarly, for $\sigma\in\SS_{n}$, the symmetric group on $n$ elements, and a graded vector space $V$, we define the map $\sigma(-):V^{\otimes n}\to V^{\otimes n}$ on homogeneous elements as $v_{1}\otimes\cdots\otimes v_{n}\mapsto\epsilon(\sigma;v_{1},\ldots,v_{n})v_{\sigma(1)}\otimes\cdots\otimes v_{\sigma(n)}$, where $\epsilon(\sigma;v_{1},\ldots,v_{n})$ is the Koszul sign.

Given a graded vector space $V$, we denote the free graded symmetric algebra on $V$ by $\operatorname{Sym}^{\bullet}(V)$, for which we let $\vee$ denote the graded-commutative product. Similarly, we denote the graded skew-symmetric algebra on a vector space $V$ by $\bigwedge^{\bullet}V$, and the graded skew-symmetric product by $\wedge$. Unless explicitly stated, a (skew-)symmetric map, is always taken to be graded (skew-)symmetric.

A (commutative) dg algebra ((c)dga) $(A,d,m)$, is a dg vector space $(A,d)$, equipped with a (symmetric) bilinear map $m:A\otimes A\to A$ such that $d$ is a derivation with respect to $m$. A semifree cdga is a freely generated cdga, that is, a cdga whose underlying graded commutative algebra is the free graded symmetric algebra on some graded vector space $V$.

Let $[n]$ be the one-dimensional graded vector space concentrated in degree $-n$. For any vector space $V$ we define $V[n]\coloneqq[n]\otimes V$, so that $(V[n])^{k}\cong V^{n+k}$ as vector spaces.

We will sometimes consider graded vector spaces equipped with an additional grading called weight. When we do, each graded piece $V^{n}$ of a graded vector space $V$ has a decomposition $V^{n}=\bigoplus_{p\in\mathbbm{Z}}V^{n,p}$. We should remark that the weight grading do not affect potential Koszul signs, rather it just provides an additional bookkeeping tool.⁶⁶6This grading convention differs from, but is equivalent to, the conventions of most literature on the pure spinor superfield formalism. In our conventions, the degree and weight are, for those well versed in the pure spinor literature, the totalised (i.e. ghost number) and (the negative of the) internal degree, respectively. Our conventions match the Tate bigrading of [1]. The aforementioned structures of dg vector spaces and (c)dgas are trivially generalised to this setting by assigning weight zero to the differential and multiplications, respectively. We adopt the convention that taking duals reverse the weight grading, thus;

A (deformation) retract of dg vector spaces $(M,d_{M})$ and $(N,d_{N})$ is a diagram

where $i$ and $p$ are morphisms of dg vector spaces, and $h$ is a degree $-1$ map such that $pi=\operatorname{\boldsymbol{1}}_{N}$ and $\operatorname{\boldsymbol{1}}_{M}-ip=d_{M}h+hd_{M}$. If moreover the side conditions $hh=hi=ph=0$ hold, the retract is called a strong (deformation) retract. Similarly, for $(M,d_{M})$ and $(N,d_{N})$ associative dg algebras, a strong (deformation) retract of associative dg algebras or a strong multiplicative retract [29], is a strong deformation retract of the underlying dg vector spaces, with the additional condition that $i$ and $p$ are algebra morphisms, and $h$ is a $(ip,\operatorname{\boldsymbol{1}})$-derivation.⁷⁷7If $(A,\cdot)$ is an associative algebra and $f,g\in\operatorname{End}(A)$, then $s\in\operatorname{End}(A)$ is called an $(f,g)$-derivation if $s(x\cdot y)=s(x)\cdot g(y)\pm f(x)\cdot s(y)$. The following result is then standard in the literature.

Given a dg vector space $(V,d)$, it is always possible [31] to define a deformation retract to the cohomology of $d$;

$L_{\infty}$-algebras are a homotopy generalisation of the notion of a Lie algebra.

We will oftentimes abbreviate algebras by just their underlying vector space, e.g. we may refer to an $L_{\infty}$-algebra $(\mathfrak{g},\mu_{k})$ as just $\mathfrak{g}$, leaving the additional structure implicit. An $L_{\infty}$-algebra $\mathfrak{g}$ is called strict if $\mu_{k}=0$ for all $k>2$, and minimal if $\mu_{1}=0$. Note that, the relations (11) imply that $(\mathfrak{g},\mu_{1})$ is a dg vector space.

An $L_{\infty}$-algebra structure on a graded vector space $\mathfrak{g}$ can equivalently be described by the datum of a degree one square-zero derivation $d_{\mathfrak{g}}$ (i.e. $d_{\mathfrak{g}}^{2}=0$) on the free algebra $C^{\bullet}(\mathfrak{g})\coloneqq\operatorname{Sym}^{\bullet}(\mathfrak{g}^{*}[-1])$, preserving the augmentation (see for example [32]). The semifree cdga $(C^{\bullet}(\mathfrak{g}),d_{\mathfrak{g}})$, is usually called the (Chevalley–Eilenberg) cochains of $\mathfrak{g}$. This correspondence between semifree cdgas and $L_{\infty}$-algebras is an instance of the Koszul duality between commutative algebras and Lie algebras.⁸⁸8The interested reader can look at [33, 34, 35, 36, 37, 38, 31, 39] for some selected works related to Koszul duality. We say that $\mathfrak{g}$ and $C^{\bullet}(\mathfrak{g})$ are Koszul dual.

A morphism of $L_{\infty}$-algebras (or $L_{\infty}$-morphism) $\Phi:\mathfrak{g}\rightsquigarrow\mathfrak{h}$ is a degree 0 morphisms of the corresponding cochains

such that $\Phi^{*}(1)=1$. The definition of a morphism of $L_{\infty}$-algebras is equivalent to a family of skew-symmetric multilinear maps

of degree $1-k$ satisfying some quadratic coherence relations involving the brackets of $\mathfrak{g}$ and $\mathfrak{h}$. In particular, $\phi^{(1)}$ is a morphism of the underlying cochain complexes. The explicit relations are not of interests to us in this paper and we refrain from writing them out, the reader is referred to [40] for more details.

We call an $L_{\infty}$-morphism an $L_{\infty}$-(quasi-)isomorphism if the first component map is a (quasi-)isomorphism of the underlying cochain complexes.

The notion of a representation of (or module over) a Lie algebra generalizes to the setting of homotopy algebras as follows. Recall that a representation of a Lie algebra $\mathfrak{g}$ on a vector space $M$ is a Lie algebra homomorphism $\rho:\mathfrak{g}\to\operatorname{End}(M)$, where $\operatorname{End}(M)$ is given a Lie algebra structure through the commutator bracket. More generally, if $(M,d_{M})$ is a dg vector space, then $\operatorname{End}(M)$ is given the structure of a dg Lie algebra, with differential $[d_{M},-]$.

The definition of an $L_{\infty}$-module is equivalent to a family of component maps (sometimes referred to as higher actions)

of degree $1-k$, which, together with the $\mu_{k}$’s, satisfy some quadratic coherence relations (see e.g. [40], and references therein). We refer to an $L_{\infty}$-module as strict if $\rho^{(k)}=0$ for $k>1$.

Similarly, to the definition of a morphism of $L_{\infty}$-algebras, a morphism $\phi:M\rightsquigarrow N$ of $L_{\infty}$-modules is a family of component maps $\phi^{(k)}:\mathfrak{g}^{\wedge k}\otimes M\to N$, satisfying some quadratic coherence relations [42]. Furthermore, if the zeroth component is a (quasi-)isomorphism of the underlying cochain complexes, $\phi$ is called a (quasi-)isomorphism of $L_{\infty}$-modules.

Given an algebra $M$ and a representation $\rho:\mathfrak{g}\to\operatorname{End}(M)$, of a Lie algebra $\mathfrak{g}$ on the underlying vector space of $M$, a suitable notion for $M$ to be an algebra over $\mathfrak{g}$ is to say that the morphism $\rho$ factors through

where $\operatorname{Der}(M)$ is the Lie algebra of derivations of the multiplication on $M$. An example of this is the adjoint representation of $\mathfrak{g}$ on itself; because of the Jacobi identity $\mathfrak{g}$ acts through derivations. Another perhaps less trivial example is to consider the action of vector fields on the algebra of functions on a manifold.

In [24], motivated by the open-closed string field theory of Zwiebach [25], Kajiura and Stasheff generalise this construction to the case when $\mathfrak{g}$ is an $L_{\infty}$-algebra and $M$ is an $A_{\infty}$-algebra. Recall that, similarly to the definition of an $L_{\infty}$-algebra, an $A_{\infty}$-algebra on a vector space $A$ is a family of degree $2-k$ multilinear maps $m_{k}:A^{\otimes k}\to A$, satisfying some quadratic coherence relations [31]. $M$ is then called an $A_{\infty}$-algebra over an $L_{\infty}$-algebra, and the way $\mathfrak{g}$ acts is through what are called homotopy derivations⁹⁹9This was later generalised to homotopy derivations of $L_{\infty}$-algebras in [44] and to algebras over an operad in [45].; that is, the higher action maps are derivations with respect to each product, only up to homotopy.

In particular, if $A$ is a strict $A_{\infty}$-algebra, with differential $d$ and multiplication $m$ (that is, an associative dga), (22) implies [24], for $q=1$,

We thus see that $\theta^{1}$ of (LABEL:eq:strictder) is a derivation of $m$ only up to homotopy, controlled by $\theta^{2}$. We will call a homotopy derivation strict if $\theta^{k}=0$ for all $k>1$.

As mentioned in [24], the definition of an $A_{\infty}$-algebra over an $L_{\infty}$-algebra encodes several substructures. In particular, the relations (25) imply that $(\mathfrak{g}\oplus A,\mu_{1}+n_{0,1})$ is a cochain complex. Furthermore, the maps $n_{0,q}$ define an $A_{\infty}$-algebra structure on $A$, and the maps $n_{p,1}$ define an $L_{\infty}$-representation of the $L_{\infty}$-algebra $\mathfrak{g}$ on the graded vector space $A$. The additional maps then correspond to higher maps of homotopy derivations, controlling the failure of the $n_{p,1}$ to act through strict derivations.

It is clear that, given a (dg) Lie algebra $\mathfrak{g}$, the above notion of an algebra over $\mathfrak{g}$ (i.e. (20)) is a special case of an $A_{\infty}$-algebra over an $L_{\infty}$-algebra. Indeed, a (dg) Lie algebra is in particular an $L_{\infty}$-algebra, and a (dg) associative algebra is a special case of an $A_{\infty}$-algebra. Moreover, having $\mathfrak{g}$ act through (strict) derivations is equivalent to an $A_{\infty}$-algebra over an $L_{\infty}$-algebra where only ($\mu_{1}$,) $\mu_{2}$, $n_{1,1}$, ($n_{0,1}$,) and $n_{1,1}$ are nonzero.