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

Action of vectorial Lie superalgebras on some split supermanifolds

Arkady Onishchik N/A
Abstract

The “curved” super Grassmannian is the supervariety of subsupervarieties of purely odd dimension kk in a supervariety of purely odd dimension nn, unlike the “usual” super Grassmannian which is the supervariety of linear subsuperspacies of purely odd dimension kk in a superspace of purely odd dimension nn. The Lie superalgebras of all and Hamiltonian vector fields on the superpoint are realized as Lie superalgebras of derivations of the structure sheaves of certain “curved” super Grassmannians.

keywords:
Lie superalgebra, homogeneous supermanifold.
\msc

Primary 17A70, 17B20, 17B70; Secondary 70F25 \VOLUME30 \NUMBER3 \DOIhttps://doi.org/10.46298/cm.10455 {paper}

Preface of the editor

The manuscript of this paper appeared as a preprint in proceedings of the “Seminar on Supersymmetries”, see http://staff.math.su.se/mleites/sos.html, and in Russian, see [52] in the list of references in the jubilee paper [AVGDZKLST*]. I updated the references; the ones I added are endowed with an asterisk. The abstract and comments are due to me. For a comprehensive description of simple Lie superalgebras of vector fields over algebraically closed fields of any characteristic, see  [BGLLS*]. D. Leites

1 Introduction

The Lie superalgebra Wn:=DerΛ[ξ1,,ξn]W_{n}:=\mathop{\mathrm{Der}}\nolimits\Lambda_{\mathbb{C}}[\xi_{1},\ldots,\xi_{n}] consisting of all vector fields on the superpoint 𝒞0,n{\mathcal{C}}^{0,n} is isomorphic, as shown in [S1], to the Lie superalgebra of vector fields, i.e., the global derivations of the structure sheaf, see [O], of a split complex supermanifold 𝒞𝒢rn1n\mathcal{CG}r_{n-1}^{n} determined by the tautological vector bundle of rank n1n-1 on the complex Grassmann manifold Grn1n\text{Gr}_{n-1}^{n}. The Lie superalgebra HnH_{n}, the subsuperalgebra of WnW_{n} consisting of Hamiltonian vector fields on the superpoint 𝒞0,n{\mathcal{C}}^{0,n}, is isomorphic to the Lie superalgebra of vector fields on a split complex supermanifold 𝒞𝒬n2\mathcal{CQ}^{n-2} associated with a vector bundle of rank n1n-1 orthogonal to the tautological bundle on the quadric Qn2Pn1Q^{n-2}\subset{\mathbb{C}}P^{n-1}. However, the method used in [S1], [S2] does not allow one to indicate explicitly these isomorphisms. In this paper I explicitly construct the WnW_{n}- and HnH_{n}-actions on the supermanifolds 𝒞𝒢rn1n\mathcal{CG}r_{n-1}^{n} and 𝒞𝒬n2\mathcal{CQ}^{n-2}; I give a new version of the proof of the above results. D. Leites told me that the supermanifolds 𝒞𝒢rn1n\mathcal{CG}r_{n-1}^{n} and 𝒞𝒬n2\mathcal{CQ}^{n-2} are the simplest examples of what Manin called “curved” Grassmannians and “curved” quadrics, see [Ma*]. They were introduced in [KL*]. (Compare with the Grassmannians of linear subsuperspaces in a linear superspace, see [Ma*, Ch.4, § 3]. For the complete list of homogeneous superdomains associated with the known Lie superalgebras of polynomial growth, see [L*]. D.L.)

2 Superization of a construction due to Serre

In this section, I superize a construction Serre introduced in [Se]. This enables us to interpret elements of a Lie superalgebra as Hamiltonian vector fields on the superpoint. Let VV be a purely odd vector space over {\mathbb{C}}, and V¯\overline{V} a second copy of the same space considered as purely even. The change of parity VV¯V\longrightarrow\overline{V} will be denoted by xx¯x\mapsto\overline{x} on every non-zero xVx\in V. Construct the Koszul complex of the {\mathbb{Z}}-graded algebra

A:=S(V¯V)=S(V¯)Λ(V){A:=S(\overline{V}\oplus V)=S(\overline{V})\otimes\Lambda(V)}

which can be naturally considered as a free supercommutative superalgebra. There exists a unique derivation dDer1Ad\in\mathop{\mathrm{Der}}\nolimits_{-1}A such that dx=x¯dx=\overline{x} and dx¯=0d\overline{x}=0 for any xVx\in V. Obviously, d2=0d^{2}=0. Consider the ZZ-graded Lie superalgebra W(V)=DerΛ(V)W(V)=\mathop{\mathrm{Der}}\nolimits\Lambda(V). Any element δW(V)\delta\in W(V) can be uniquely extended to a derivation δ~DerA\tilde{\delta}\in\mathop{\mathrm{Der}}\nolimits A such that [δ~,d]=0[\tilde{\delta},d]=0, see [K]. The correspondence δδ~\delta\mapsto\tilde{\delta} is a faithful linear representation of the Lie superalgebra W(V)W(V) in AA. Let ωS2(V¯)\omega\in S^{2}(\overline{V}) be a nondegenerate bilinear form. Set

H(ω):={δW(V)δ~(ω)=0}.H(\omega):=\{\delta\in W(V)\mid\tilde{\delta}(\omega)=0\}.

Then H(ω)H(\omega) is a {\mathbb{Z}}-graded subalgebra in W(V)W(V) called the Lie superalgebra of Hamiltonian vector fields. Set

DH(ω):={δW(V)δ~(ω)=φω for some φA}.DH(\omega):=\{\delta\in W(V)\mid\tilde{\delta}(\omega)=\varphi\omega\text{ for some }\varphi\in A\}.

Clearly, DH(ω)DH(\omega) is a {\mathbb{Z}}-graded subalgebra of W(V)W(V), and H(ω)H(\omega) is its ideal. Hereafter we assume that dimV=n\dim V=n. Set Wn:=W(V)W_{n}:=W(V), Hn:=H(ω){H_{n}:=H(\omega)}, DHn:=DH(ω)DH_{n}:=DH(\omega) since these algebras are determined, up to an isomorphism, by dimV\dim V. In VV, select a basis ξ1,,ξn\xi_{1},\ldots,\xi_{n}. Then the elements xi=ξ¯i=dξix_{i}=\overline{\xi}_{i}=d\xi_{i}, where i=1,,ni=1,\ldots,n, constitute a basis in V¯\overline{V} and

A=[x1,,xn]Λ[ξ1,,ξn].A={\mathbb{C}}[x_{1},\ldots,x_{n}]\otimes\Lambda[\xi_{1},\ldots,\xi_{n}].

Obviously, d=1inxiξid=\sum\limits_{1\leqslant i\leqslant n}x_{i}\partial_{\xi_{i}}. Notice that ξiDer1A\partial_{\xi_{i}}\in\mathop{\mathrm{Der}}\nolimits_{-1}A coincides with the extension ~ξi\tilde{\partial}_{\xi_{i}} of ξiW(V)1\partial_{\xi_{i}}\in W(V)_{-1}. {Lemma} 1) If δ~(ω)=φω\tilde{\delta}(\omega)=\varphi\omega for some δW(V)\delta\in W(V), φA\varphi\in A, then φ\varphi\in{\mathbb{C}}. 2) There is the following semidirect sum decomposition:

DH(ω)=H(ω)E,whereE=1inξiξi.DH(\omega)=H(\omega)\ltimes{\mathbb{C}}E,\quad\text{where}\quad E=\sum\limits_{1\leqslant i\leqslant n}\xi_{i}\partial_{\xi_{i}}.
Proof.

1) We may assume that (xi)i=1n(x_{i})_{i=1}^{n} is a basis in which ω\omega is the form

ω=1inxi2=1in(dξi)2.\omega=\sum\limits_{1\leqslant i\leqslant n}x_{i}^{2}=\sum\limits_{1\leqslant i\leqslant n}(d\xi_{i})^{2}.

Then we have

δ~ω=21inxiδ~xi=(1)k21inxid(δξi)for anyδW(V)k.\tilde{\delta}\omega=2\sum\limits_{1\leqslant i\leqslant n}x_{i}\tilde{\delta}x_{i}=(-1)^{k}2\sum\limits_{1\leqslant i\leqslant n}x_{i}d(\delta\xi_{i})\quad\text{for any}\quad\delta\in W(V)_{k}.

Setting hi=δξih_{i}=\delta\xi_{i} we get

δ~ω=(1)k21i,jnxixjhjξi,wherehjξiΛk(V).\tilde{\delta}\omega=(-1)^{k}2\sum\limits_{1\leqslant i,j\leqslant n}x_{i}x_{j}\frac{\partial h_{j}}{\partial\xi_{i}},\quad\text{where}\quad\frac{\partial h_{j}}{\partial\xi_{i}}\in\Lambda^{k}(V).

If δ\delta satisfies the conditions of Lemma 2, then φ=(1)k2hiξi\varphi=(-1)^{k}2\frac{\partial h_{i}}{\partial\xi_{i}} for any i=1,,ki=1,\ldots,k. In particular, φΛk(V)\varphi\in\Lambda^{k}(V). Further, for any ii, we have φξi=(1)k22hiξi2=0\frac{\partial\varphi}{\partial\xi_{i}}=(-1)^{k}2\frac{\partial^{2}h_{i}}{\partial\xi_{i}^{2}}=0. Therefore, φ\varphi\in{\mathbb{C}}. 2) Observe that E(ω)=2ωE(\omega)=2\omega, and so EDH(ω)0E\in DH(\omega)_{0}. Further, if δDH(ω)\delta\in DH(\omega), then, due to the proved above, δ~ω=cω\tilde{\delta}\omega=c\omega, where c{c\in{\mathbb{C}}}. Hence, (δ~12cE)(ω)=0\left(\tilde{\delta}-\frac{1}{2}\,cE\right)(\omega)=0, i.e., δ=δ0+12cE\delta=\delta_{0}+\frac{1}{2}\,cE, where δ0H(ω){\delta_{0}\in H(\omega)}. ∎

3 Vector bundles over Pn1{\mathbb{C}}P^{n-1} and Qn2Q^{n-2} and supermanifolds

Define some special supermanifolds associated with vector bundles over Qn2Pn1Q^{n-2}\subset{\mathbb{C}}P^{n-1} and Pn1{\mathbb{C}}P^{n-1}. Let dimV=n\dim V=n, and P(V)P(V) the corresponding protective space. Let us assume that the nonzero elements of VV^{*} are odd and those of V¯\overline{V}{}^{*} are even. In VV, select a basis e1,,ene_{1},\ldots,e_{n}, and consider the dual bases ξ1,,ξn\xi_{1},\ldots,\xi_{n} and x1,,xnx_{1},\ldots,x_{n} of VV^{*} and V¯\overline{V}{}^{*}, respectively. In the notation of § 2 (applied to VV^{*} and V¯\overline{V}^{*}) we have xi=ξ¯ix_{i}=\overline{\xi}_{i}. The elements x1,,xnx_{1},\ldots,x_{n} are homogeneous coordinates on P(V)P(V); this means that the stalk za{\mathcal{F}}_{z}^{a} of the structure sheaf a{\mathcal{F}}^{a} of the algebraic variety P(V)P(V) at zP(V)z\in P(V) is a subring of the field (V)=(x1,,xn){\mathbb{C}}(V)={\mathbb{C}}(x_{1},\ldots,x_{n}) consisting of elements of the form f/gf/g, where f,gf,g are homogeneous polynomials of the same degree in [x1,,xn]=S(V¯){\mathbb{C}}[x_{1},\ldots,x_{n}]=S(\overline{V}^{*}) and g(z)0g(z)\neq 0. Consider the trivial vector bundle P(V)×VP(V)\times V^{*} and its subbundle EP(V)×VE\subset P(V)\times V^{*} consisting of the pairs (x,y)({\mathbb{C}}x,y), where xV{0}x\in V\setminus\{0\} and yAnnx={αVα(x)=0}y\in\mathop{\mathrm{Ann}}\nolimits x=\{\alpha\in V^{*}\mid\alpha(x)=0\}. Clearly, EE is an algebraic vector bundle of rank n1n-1 over P(V)P(V) with the fiber

Ex=Annx, where xV{0}).E_{{\mathbb{C}}x}=\mathop{\mathrm{Ann}}\nolimits x,\text{~{}~{} where $x\in V\setminus\{0\})$.}

The map xAnnx{\mathbb{C}}x\mapsto\mathop{\mathrm{Ann}}\nolimits x identifies P(V)P(V) with the Grassmann variety Grn1(V)\text{Gr}_{n-1}(V^{*}), which consists of (n1)(n-1)-dimensional subspaces in VV^{*}, and EE is the tautological bundle over this Grassmann variety. Let aaV{\mathcal{E}}^{a}\subset{\mathcal{F}}^{a}\otimes V^{*} be the sheaf of germs of the polynomial sections of EE. The variety P(V)P(V) can be endowed with two structures of a split complex algebraic supervariety: one is determined by the sheaf 𝒪^=aaΛ(V){\hat{\mathcal{O}}{}^{a}={\mathcal{F}}^{a}\otimes\Lambda(V^{*})}, and the other by its subsheaf 𝒪a=Λ(a){\mathcal{O}}^{a}=\Lambda({\mathcal{E}}^{a}). Besides, 𝒪^=Λ(V)\hat{\mathcal{O}}={\mathcal{F}}\otimes\Lambda(V^{*}) and 𝒪=Λ()𝒪^{\mathcal{O}}=\Lambda({\mathcal{E}})\subset\hat{\mathcal{O}}, where {\mathcal{F}} is the sheaf of holomorphic functions on P(V)P(V) and {\mathcal{E}} is the sheaf of germs of holomorphic sections of EE, determine two structures of a split complex analytic supermanifold associated with P(V)P(V). Consider the superalgebra

A:=S(V¯)Λ(V)=[x1,,xn]Λ[ξ1,,ξn]A:=S(\overline{V}{}^{*})\otimes\Lambda(V^{*})={\mathbb{C}}[x_{1},\ldots,x_{n}]\otimes\Lambda[\xi_{1},\ldots,\xi_{n}]

and the derivation dDer1Ad\in\mathop{\mathrm{Der}}\nolimits_{-1}A constructed in § 2 (with VV replaced by VV^{*} in these constructions). Let

B:=(V)Λ(V)=(x1,,xn)Λ[ξ1,,ξn]B:={\mathbb{C}}(V)\otimes\Lambda(V^{*})={\mathbb{C}}(x_{1},\ldots,x_{n})\otimes\Lambda[\xi_{1},\ldots,\xi_{n}]

be the localization of AA with respect to the multiplicative system S(V¯){0}S(\overline{V}{}^{*})\setminus\{0\}. The algebra BB has a natural supercommutative superalgebra structure, and dd can be uniquely extended to an odd derivation of BB, which we will denote also by dd. Clearly,

𝒪^=zazaΛ(V)B for any zP(V).\hat{\mathcal{O}}{}_{z}^{a}={\mathcal{F}}_{z}^{a}\otimes\Lambda(V^{*})\subset B\text{~{}~{} for any $z\in P(V)$.}
{Lemma}

We have 𝒪za=𝒪^zaKerD{\mathcal{O}}_{z}^{a}=\hat{\mathcal{O}}{}_{z}^{a}\cap\mathop{\mathrm{Ker}}\nolimits D for any zP(V)z\in P(V).

Proof.

For any xV{0}x\in V\setminus\{0\}, consider the derivation

dx=1inxiξiDer1Λ(V)d_{x}=\sum\limits_{1\leqslant i\leqslant n}x_{i}\partial_{\xi_{i}}\in\mathop{\mathrm{Der}}\nolimits_{-1}\Lambda(V^{*})

uniquely determined by the condition dx(ξ)=ξ(x)d_{x}(\xi)=\xi(x) for any ξV\xi\in V^{*}. Clearly,

Kerdx=Λ(Annx)=Λ(Ex).\mathop{\mathrm{Ker}}\nolimits d_{x}=\Lambda(\mathop{\mathrm{Ann}}\nolimits x)=\Lambda(E_{{\mathbb{C}}x}).

Let u:=1irφiviBu:=\sum\limits_{1\leqslant i\leqslant r}\varphi_{i}v_{i}\in B, where φia\varphi_{i}\in{\mathcal{F}}^{a} and viΛ(V)v_{i}\in\Lambda(V^{*}). Considering uu as a function on V{0}V\setminus\{0\} with values in Λ(V)\Lambda(V^{*}) we see that u(x)=1irφi(x)vi{u(x)=\sum\limits_{1\leqslant i\leqslant r}\varphi_{i}({\mathbb{C}}x)v_{i}} at any point xx from its domain implying

dxu(x)=1irφi(x)dxvi=1irφi(x)1jnxjviξj.d_{x}u(x)=\sum\limits_{1\leqslant i\leqslant r}\varphi_{i}({\mathbb{C}}x)d_{x}v_{i}=\sum\limits_{1\leqslant i\leqslant r}\varphi_{i}({\mathbb{C}}x)\sum\limits_{1\leqslant j\leqslant n}x_{j}\frac{\partial v_{i}}{\partial\xi_{j}}.

On the other hand,

du=1irφidvi=φi1inxjviξj.du=\sum\limits_{1\leqslant i\leqslant r}\varphi_{i}dv_{i}=\sum\limits\varphi_{i}\sum\limits_{1\leqslant i\leqslant n}x_{j}\frac{\partial v_{i}}{\partial\xi_{j}}.

Therefore, (du)(x)=dxu(x)(du)(x)=d_{x}u(x) on a Zariski open subset of VV. Hence, du=0du=0 if and only if u(x)Λ(Ex)u(x)\in\Lambda(E_{{\mathbb{C}}x}) for all xVx\in V from the domain of uu. ∎

Now, define a subsupermanifold of (P(V),𝒪a)(P(V),{\mathcal{O}}^{a}) whose underlying manifold is the quadric QP(V)Q\subset P(V). Let ω\omega be a non-degenerate quadratic function on VV; it can be considered as an element of S2(V¯)S^{2}(\overline{V}{}^{*}). Let QQ be the quadric in P(V)P(V) given by the equation ω=0\omega=0 and E=E|QE^{\prime}=E|_{Q} the restriction of EE to QQ. If we identify VV^{*} with VV with the help of the non-degenerate symmetric bilinear form corresponding to ω\omega, then EE^{\prime} is identified with the subbundle of Q×VQ\times V orthogonal to the tautological line bundle over QQ. Let a{\mathcal{F}}^{\prime a} and a{\mathcal{E}}^{\prime a} (resp. {\mathcal{F}}^{\prime} and {\mathcal{E}}^{\prime}) be the sheaves of polynomial (resp. holomorphic) functions on QQ and polynomial (resp. holomorphic) sections of {\mathcal{E}}^{\prime}, respectively. Set 𝒪a:=Λ(a){{\mathcal{O}}^{\prime a}:=\Lambda({\mathcal{E}}^{\prime a}}) and 𝒪:=Λ(){\mathcal{O}}^{\prime}:=\Lambda({\mathcal{E}}^{\prime}). Let us give a description of 𝒪a{\mathcal{O}}^{\prime a} similar to the above description of 𝒪a{\mathcal{O}}^{a}. For this, consider the superalgebra

A:=A/ωA=[Q^]Λ(V),A^{\prime}:=A/\omega A={\mathbb{C}}[\hat{Q}]\otimes\Lambda(V^{*}),

where [Q^]:=S(V¯)/ωS(V¯){\mathbb{C}}[\hat{Q}]:=S(\overline{V}{}^{*})/\omega S(\overline{V}{}^{*}) is the algebra of polynomial functions on the cone Q^V\hat{Q}\subset V given by the equation ω=0\omega=0; the localization of AA^{\prime} is B:=(Q^)Λ(V){B^{\prime}:={\mathbb{C}}(\hat{Q})\otimes\Lambda(V^{*})}. The algebras za{\mathcal{F}}{}^{\prime a}_{z} and 𝒪za{\mathcal{O}}_{z}^{\prime a}, where zQz\in Q, are embedded into BB^{\prime}. The derivation dd transforms ωA\omega A into itself, and therefore determines an odd derivation dd^{\prime} of AA^{\prime} and BB^{\prime}. Lemma 3 implies that

𝒪za=(zaΛ(V))Kerdfor anyzQ.{\mathcal{O}}_{z}^{\prime a}=({\mathcal{F}}_{z}^{\prime a}\otimes\Lambda(V^{*}))\cap\mathop{\mathrm{Ker}}\nolimits d^{\prime}\quad\text{for any}\quad z\in Q.

4 Several remarks on vector fields on algebraic and analytic supervarieties

Let MM be a nonsingular complex algebraic variety, EE an algebraic vector bundle over MM. Denote by a{\mathcal{F}}^{a}, 𝒯a{\mathcal{T}}^{a}, a{\mathcal{E}}^{a} the structure sheaf on MM, the tangent sheaf on MM, and the locally free algebraic sheaf corresponding to EE, respectively. We denote by the same letters without the superscript aa the corresponding analytic sheaves on MM. In particular, (M,)(M,{\mathcal{F}}) is the complex analytic manifold corresponding to the algebraic variety MM. The sheaves 𝒪a=Λa(a){\mathcal{O}}^{a}=\Lambda_{{\mathcal{F}}^{a}}({\mathcal{E}}^{a}) and 𝒪=Λ(){\mathcal{O}}=\Lambda_{\mathcal{F}}({\mathcal{E}}) rig MM with structures of a split algebraic and a split analytic supervariety, respectively. We call the sheaves of {\mathbb{Z}}-graded Lie superalgebras Der𝒪a\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}^{a} and Der𝒪\mathop{\mathrm{Der}}\nolimits{\mathcal{O}} the sheaves of vector fields on these supervarieties. {Lemma} There exists a natural injective homomorphism of sheaves of {\mathbb{Z}}-graded Lie superalgabras Der𝒪aDer𝒪\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}^{a}\longrightarrow\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}.

Proof.

As shown in [O], for any kk\in{\mathbb{Z}}, every γDerk𝒪\gamma\in\mathop{\mathrm{Der}}\nolimits_{k}{\mathcal{O}} can be identified with a pair (γ0,γ1)(\gamma_{0},\gamma_{1}), where

γ0Hom(,Λk+1())\gamma_{0}\in\mathop{\mathrm{Hom}}\nolimits_{\mathcal{F}}({\mathcal{E}},\Lambda^{k+1}({\mathcal{E}})) and γ1Hom(,Λk())=𝒯Λk()\gamma_{1}\in\mathop{\mathrm{Hom}}\nolimits_{\mathbb{C}}({\mathcal{F}},\Lambda^{k}({\mathcal{E}}))={\mathcal{T}}\mskip 4.0mu plus 2.0mu minus 4.0mu{\otimes}_{\mathcal{F}}\mskip 4.0mu plus 2.0mu minus 4.0mu\Lambda^{k}({\mathcal{E}}) (1)

such that

γ0(φs)=γ1(φ)s+φγ0(s)γ1(φψ)=γ1(φ)ψ+φγ1(ψ)for any s and φ,ψ.\begin{array}[]{l}\gamma_{0}(\varphi s)=\gamma_{1}(\varphi)s+\varphi\gamma_{0}(s)\\ \gamma_{1}(\varphi\psi)=\gamma_{1}(\varphi)\psi+\varphi\gamma_{1}(\psi)\end{array}\qquad\text{for any }s\in{\mathcal{E}}\text{ and }\varphi,\psi\in{\mathcal{F}}. (2)

A similar statement holds also for Der𝒪a\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}^{a}. Since the sections of the sheaves

Homa(a,Λk+1(a))\mathop{\mathrm{Hom}}\nolimits_{{\mathcal{F}}^{a}}({\mathcal{E}}^{a},\Lambda^{k+1}({\mathcal{E}}^{a})) (resp. Hom(,Λk+1())\mathop{\mathrm{Hom}}\nolimits_{\mathcal{F}}({\mathcal{E}},\Lambda^{k+1}({\mathcal{E}})))

are algebraic (resp. holomorphic) homomorphisms of vector bundles EΛk+1(E){E\longrightarrow\Lambda^{k+1}(E)}, there is a natural embedding

Homa(a,Λk+1(a))Hom(,Λk+1()).\mathop{\mathrm{Hom}}\nolimits_{{\mathcal{F}}^{a}}({\mathcal{E}}^{a},\Lambda^{k+1}({\mathcal{E}}^{a}))\longrightarrow\mathop{\mathrm{Hom}}\nolimits_{\mathcal{F}}({\mathcal{E}},\Lambda^{k+1}({\mathcal{E}})).

Further, the natural embedding 𝒯a𝒯{\mathcal{T}}^{a}\longrightarrow{\mathcal{T}} induces the embedding

𝒯aaΛk(a)𝒯Λk().{\mathcal{T}}^{a}\mskip 4.0mu plus 2.0mu minus 4.0mu{\otimes}_{{\mathcal{F}}^{a}}\mskip 4.0mu plus 2.0mu minus 4.0mu\Lambda^{k}({\mathcal{E}}^{a})\longrightarrow{\mathcal{T}}\mskip 4.0mu plus 2.0mu minus 4.0mu{\otimes}_{\mathcal{F}}\mskip 4.0mu plus 2.0mu minus 4.0mu\Lambda^{k}({\mathcal{E}}).

Therefore, to every pair (γ0,γ1)Homa(a,Λk+1(a))×Hom(a,Λk(a))(\gamma_{0},\gamma_{1})\in\mathop{\mathrm{Hom}}\nolimits_{{\mathcal{F}}^{a}}({\mathcal{E}}^{a},\Lambda^{k+1}({\mathcal{E}}^{a}))\times\mathop{\mathrm{Hom}}\nolimits_{\mathbb{C}}({\mathcal{F}}^{a},\Lambda^{k}({\mathcal{E}}^{a})) that satisfies conditions similar to (2), there corresponds a pair, see (1), satisfying (2), as is easy to verify. This is the desired embedding Der𝒪aDer𝒪{\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}^{a}\longrightarrow\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}}. ∎

{Corollary}

There exists an injective homomorphism of {\mathbb{Z}}-graded Lie superalgebras of vector fields 𝔡a𝔡{\mathfrak{d}}^{a}\longrightarrow{\mathfrak{d}}, where 𝔡a=Γ(M,Der𝒪a){\mathfrak{d}}^{a}=\Gamma(M,\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}^{a}) and 𝔡=Γ(M,Der𝒪){\mathfrak{d}}=\Gamma(M,\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}). The results of [Se] imply that the homomorphism 𝔡a𝔡{\mathfrak{d}}^{a}\longrightarrow{\mathfrak{d}} is an isomorphism for any projective variety MM.

5 WnW_{n} and DHnDH_{n} as vectorial Lie superalgebras

Now we are able to determine the action of WnW_{n} and DHnDH_{n} on the supermanifolds constructed in § 3. Retaining the notation of § 3 let us first prove the following statement. {Lemma} Let γDerB\gamma\in\mathop{\mathrm{Der}}\nolimits B be such that γVΛ(V)\gamma V\subset\Lambda(V), so that

γxi=1invijxj{\gamma x_{i}=\sum\limits_{1\leqslant i\leqslant n}v_{ij}x_{j}}, where i=1,,ni=1,\ldots,n and vijΛ(V)v_{ij}\in\Lambda(V).

Then, γ(𝒪^)za𝒪^za\gamma(\hat{\mathcal{O}}{}_{z}^{a})\subset\hat{\mathcal{O}}{}_{z}^{a} for all zP(V)z\in P(V).

Proof.

Let u=φvu=\varphi v, where φza\varphi\in{\mathcal{F}}_{z}^{a} and vΛ(V)v\in\Lambda(V). Then,

γ(u)=γ(φ)v+φγ(v).\gamma(u)=\gamma(\varphi)v+\varphi\gamma(v).

Therefore, it suffices to verify that γ(φ)𝒪^za\gamma(\varphi)\in\hat{{\mathcal{O}}}{}_{z}^{a}. Let us express φ\varphi in the form f/gf/g, where f,g[x1,,xn]f,g\in{\mathbb{C}}[x_{1},\ldots,x_{n}] are homogeneous polynomials of the same degree and g(z)0g(z)\neq 0. Our statement follows easily from the hypothesis on γ\gamma and the identity

γ(φ)=1g2(gγ(f)fγ(g)).\gamma(\varphi)=\frac{1}{g^{2}}\,(g\gamma(f)-f\gamma(g)).

Now, let δWn=W(V)\delta\in W_{n}=W(V^{*}). As we have seen in § 2, δ\delta can be uniquely extended to a derivation δ~\tilde{\delta} of A=S(V¯)Λ(V)A=S(\overline{V}{}^{*})\otimes\Lambda(V^{*}) such that [δ~,d]=0[\tilde{\delta},d]=0. Obviously, the action of δ~\tilde{\delta} can be uniquely extended to the localization BB of AA, so (see Proof of Lemma 2)

δ~xi=δ~dξi=dδξi=±1jnxjhiξj,wherehi=δξiΛ(V).\tilde{\delta}x_{i}=\tilde{\delta}d\xi_{i}=d\delta\xi_{i}=\pm\sum\limits_{1\leqslant j\leqslant n}x_{j}\frac{\partial h_{i}}{\partial\xi_{j}},\quad\text{where}\quad h_{i}=\delta\xi_{i}\in\Lambda(V).

By Lemma 5 δ~\tilde{\delta} transforms all the algebras 𝒪^za\hat{\mathcal{O}}{}_{z}^{a}, where zP(V)z\in P(V), into themselves. Since δ~\tilde{\delta} transforms Kerd\mathop{\mathrm{Ker}}\nolimits d into itself, then Lemma 3 implies that δ(𝒪za)𝒪za\delta({\mathcal{O}}_{z}^{a})\subset{\mathcal{O}}_{z}^{a} for all zP(V)z\in P(V). Therefore, δ\delta determines a global derivation of 𝒪a{\mathcal{O}}^{a}. We have constructed a map

Wn𝔡a=Γ(P(V),Der𝒪a).W_{n}\longrightarrow{\mathfrak{d}}^{a}=\Gamma(P(V),\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}^{a}).

As is easy to see, theis map is an injective homomorphism of {\mathbb{Z}}-graded Lie superalgebras. By Lemma 4 we also have an injective homomorphism Wn𝔡=Γ(P(V),Der𝒪)W_{n}\longrightarrow{\mathfrak{d}}=\Gamma(P(V),\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}). ∎

We similarly construct an injective homomorphism

DHn𝔡=Γ(Q,Der𝒪).DH_{n}\longrightarrow{\mathfrak{d}}^{\prime}=\Gamma(Q,\mathop{\mathrm{Der}}\nolimits{\mathcal{O}}^{\prime}).

If δDHn{\delta\in DH_{n}}, then δ~DerA\tilde{\delta}\in\mathop{\mathrm{Der}}\nolimits A transforms ωA\omega A into itself, and therefore determines a derivation of AA^{\prime} from § 3 which extends to BB^{\prime} and yields a derivation of 𝒪a{\mathcal{O}}^{\prime a}. {Theorem} The above-constructed homomorphisms Wn𝔡W_{n}\longrightarrow{\mathfrak{d}} and DHn𝔡DH_{n}\longrightarrow{\mathfrak{d}}^{\prime} are isomorphisms if n2n\geqslant 2 and n5n\geqslant 5, respectively.

Proof.

By the proved above we may assume that the finite-dimensional {\mathbb{Z}}-graded Lie superalgebras 𝔡{\mathfrak{d}} and 𝔡{\mathfrak{d}}^{\prime} contain WnW_{n} and DHnDH_{n}, respectively, as subalgebras. Therefore, it suffices to prove that 𝔡{\mathfrak{d}} and 𝔡{\mathfrak{d}}^{\prime} are transitive and irreducible and 𝔡k=(Wn)k{\mathfrak{d}}_{k}=(W_{n})_{k} and 𝔡k=(DHn)k{\mathfrak{d}}^{\prime}_{k}=(DH_{n})_{k} for k=1,0k=-1,0 (see [K, Theorem 4]). A proof of these statements is contained in [S1], [S2]. This proof essentially depends on Lemma 4.1.1 in [O] and actually reduces to the calculation of 𝔡0{\mathfrak{d}}_{0} and 𝔡0{\mathfrak{d}}_{0}^{\prime} and also 𝔡0{\mathfrak{d}}_{0}-module 𝔡1{\mathfrak{d}}_{-1} and 𝔡0{\mathfrak{d}}_{0}^{\prime}-module 𝔡1{\mathfrak{d}}_{-1}^{\prime} with the help of Bott’s theorem. ∎

References

  • [AVGDZKLST*] D. N. Akhiezer, È. B. Vinberg, V. V. Gorbatsevich, V. G. Durnev, R. Zulanke, L. S. Kazarin, D. A. Leites, V. V. Serganova, V. M. Tikhomirov, Arkadii L’vovich Onishchik (on his 70th birthday). Russian Math. Surveys, 58:6 (2003), 1245–1253.
  • [BGLLS*] Bouarroudj S., Grozman P., Lebedev A., Leites D., Shchepochkina I., Simple vectorial Lie algebras in characteristic 22 and their superizations. Symmetry, Integrability and Geometry: Methods and Applications (SIGMA), 16 (2020), 089, 101 pages; arXiv:1510.07255.
  • [K] Kac V. G., Lie superalgebras. Adv. Math. 1977. 26, No. 1, 8–96.
  • [KL*] Kirillova R. Yu., Leites D. A., Instantons with gauge supergroup. In: Problems of nuclear physics and cosmic rays, 1985, no. 24 (D. V. Volkov Festschrift), Kharkov Univ. Press, Kharkov, 34–40 (in Russian) MR 88h:53032.
  • [L*] Leites D. A., On unconventional integration on supermanifolds and cross ratio on classical superspaces. In: E. Ivanov, S. Krivonos, J. Lukierski, A. Pashnev (eds.) Proceedings of the International Workshop “Supersymmetries and Quantum Symmetries”, September 21–25, 2001, Karpacz, Poland. JINR, Dubna, 2002, 251–262; arXiv:math.RT/0202194.
  • [Ma*] Manin Yu., Gauge field theory and complex geometry. Second edition. Springer-Verlag, Berlin, 1997. xii+346 pp.
  • [O] Onischik A. L., Transitive Lie superalgebras of vector fields. Communications in Mathematics 30 (2022), no. 3, 25–40.
  • [OS1*] Onishchik A. L., Serov A. A., Lie superalgebras of vector fields on splittable flag supermanifolds, Dokl. Akad. Nauk SSSR 300 (1988), no. 2, 284-287 (in Russian); English translation: Soviet Math. Dokl. 37 (1988), no.3, 652–655. MR 89i:32067.
  • [OS2*] Onishchik, A. L.; Serov, A. A. Holomorphic vector fields on super-Grassmannians. In: Lie groups, their discrete subgroups, and invariant theory. Adv. Soviet Math., 8, Amer. Math. Soc., Providence, RI, 1992, 113–129.
  • [S1] Serov A. A., Vector fields on split supermanifolds, 22–81. In: Leites D. (ed.) “Seminar on Supersymmetries”, 20/1987-26, see http://staff.math.su.se/mleites/sos.html. For a summary, see [OS1*], [OS2*].
  • [S2] Serov A. A., A realization of the Hamiltonian Lie superalgebra. In: Onishchik A. (ed.) Problems in group theory and homological algebra, 164–167, Matematika, Yaroslav. Gos. Univ., Yaroslavl, 164–167 (in Russian) MR1174807.
  • [Se] Serre J.-P., Géometrie algébraique et géometrie analytique. Ann. Inst. Fourier. 1956. 6, 1–42.