Journal of Geometry and Physics 62 (2012) 2259–2278

Contents lists available at SciVerse ScienceDirect

Journal of Geometry and Physics

journal homepage: www.elsevier.com/locate/jgp

Sheaves of N = 2 supersymmetric vertex algebras on Poisson manifolds

Joel Ekstrand a,∗ , Reimundo Heluani b , Maxim Zabzine a,c
a
Department of Physics and Astronomy, Uppsala University, Box 516, SE-751 20 Uppsala, Sweden
b
IMPA, Rio de Janeiro, RJ 22460-320, Brazil
c
Kavli Institute for Theoretical Physics, University of California, Santa Barbara, CA 93106, USA

article info abstract

Article history: We construct a sheaf of N = 2 vertex algebras naturally associated to any Poisson manifold.
Received 27 September 2011 The relation of this sheaf to the chiral de Rham complex is discussed. We reprove the result
Received in revised form 19 June 2012 about the existence of two commuting N = 2 superconformal structures on the space of
Accepted 8 July 2012
sections of the chiral de Rham complex of a Calabi–Yau manifold, but now calculated in a
Available online 15 July 2012
manifest N = 2 formalism. We discuss how the semi-classical limit of this sheaf of N = 2
vertex algebras is related to the classical supersymmetric non-linear sigma model.
Keywords:
Vertex algebra
© 2012 Elsevier B.V. All rights reserved.
SUSY vertex algebra
Poisson vertex algebra
Poisson geometry

1. Introduction

To any smooth manifold M one can associate a sheaf of vertex algebras [1], which is called the chiral de Rham complex
(CDR) of M. Locally on a d-dimensional manifold M one attaches d copies of the free bosonic βγ -system tensored with d
copies of the free fermionic bc-system. These local models are then glued along intersections of the corresponding patches
on M using appropriate automorphisms of these free field systems. One can combine these 4d fields into 2d N = 1 superfields
to obtain a sheaf of N = 1 SUSY vertex algebras [2]. More generally, one can construct a sheaf of N = 1 SUSY vertex algebras
associated to any Courant algebroid E over M, and this can be done in a coordinate free fashion [3]. Geometric properties
of M are reflected in algebraic properties of CDR. For example, if M admits a generalized Calabi–Yau structure, then there
exists an embedding of the N = 2 superconformal vertex algebra into global sections of this sheaf [4]. The reader may find
more results along these lines in [2,3,5].
There is a quasiclassical version of CDR as a sheaf of Poisson vertex algebras [6]. This can be naturally related to the
Hamiltonian treatment of supersymmetric classical non-linear sigma models [7]. Indeed, CDR can be interpreted as a formal
quantization of the non-linear sigma model. By ‘‘formal’’ we mean here that instead of working with the actual loop space
of M, one deals with the space of formal loops into M [8]. Nevertheless, the relation to sigma models is quite inspiring; e.g.
see [5].
In this note, we present a very simple construction of a sheaf of N = 2 SUSY vertex algebras on any Poisson manifold M.
We discuss the relation of this N = 2 sheaf to CDR as a sheaf of N = 1 SUSY vertex algebras. We recover the main result
from [3], about the existence of two commuting N = 2 superconformal structures on the space of sections of CDR in the
Calabi–Yau case, but now calculated in a manifest N = 2 formalism. We also briefly discuss the semiclassical limit of this
N = 2 sheaf and its relation to the Hamiltonian treatment of N = (2, 2) supersymmetric sigma models with a Kähler target.

∗ Corresponding author.
E-mail address: [email protected] (J. Ekstrand).

0393-0440/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomphys.2012.07.003
2260 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

The paper is organized as follows. In Section 2, we introduce basic notations on Poisson manifolds and supermanifolds.
In Section 3, the definition of an NK = 2 SUSY vertex algebra is given. In Section 4, we construct the sheaf of N = 2 SUSY
vertex algebras on any Poisson manifold. In Section 5, we discuss the case of a symplectic manifold. Section 6 deals with the
case of a Calabi–Yau manifold. In Section 7, we consider the semiclassical limit of the N = 2 sheaf and discuss the relation to
the N = (2, 2) supersymmetric sigma model. Section 8 contains a summary of the results in this article and a discussion of
open questions. All technical calculations are collected in Appendices. For the reader’s convenience, we collect the rules for
Λ-brackets in Appendix A. Appendix B contains the calculation for the symplectic case. Appendix C presents the proof of the
existence of an embedding of the N = (2, 2) superconformal algebra in the Calabi–Yau case, in manifest N = 2 formalism.
Appendix D contains the details of the Hamiltonian treatment of the N = (2, 2) sigma model with a Kähler target.

2. Preliminaries

In this short section, we introduce the basics of Poisson and supergeometry in order to collect notations that will be
useful below.

2.1. Poisson manifolds

Definition 1. A commutative associative algebra A over a field F is called a Poisson algebra if it is equipped with an F-bilinear
Lie bracket {, } : A ⊗F A → A satisfying the Leibniz identity:
{ab, c } = a{b, c } + {a, c }b, ∀a, b, c ∈ A.

Example 1. The trivial Poisson structure is defined by the bracket {, } = 0 over any commutative associative algebra A.

Example 2. The symplectic plane is defined as the algebra A = F[q, p] with the bracket {p, q} = 1 extended by Leibniz. We
easily find
∂f ∂g ∂f ∂g
{f , g } = − , ∀f , g ∈ A.
∂p ∂q ∂q ∂p

Example 3. Let g be a finite dimensional Lie algebra over F and g∗ its dual. Then the algebra S g of polynomials in g∗ naturally
is a Poisson algebra with the bracket
{f , g }(ζ ) := ζ [df (ζ ), dg (ζ )] , f , g ∈ S g, ζ ∈ g∗ .
 

Example 4. Example 2 can be generalized as follows. Recall that a symplectic manifold is a smooth manifold endowed with
a closed nondegenerate 2-form ω. Let A := C ∞ (M ) be the algebra of smooth functions. For each f ∈ A let Xf be the smooth
vector field defined by df (Y ) = ω(Y , Xf ). Setting
{f , g } = ω(Xf , Xg ),
we obtain a Poisson algebra structure on A.
Darboux’s theorem asserts that locally every symplectic structure is the same: for each point x in a symplectic manifold
M there exists a local system of coordinates {pi , qi }ni=1 such that in that coordinate neighborhood we have ω = i dpi ∧ dqi .
The Poisson algebra in this case looks as the tensor product of n copies of Example 2.

Definition 2. A Poisson manifold is a smooth manifold M where C ∞ (M ) is a Poisson algebra.

It follows from the Leibniz identity that for any function f there exists a smooth vector field (called Hamiltonian) Xf such
that {f , g } = dg (Xf ). Thus, there exists a bivector field Π (called the Poisson bivector) defined by

{f , g } = df ⊗ dg (Π ).
 
(2.1)
Conversely, given a bivector Π , one can define a skew-symmetric bilinear bracket by (2.1). This operation will satisfy the
Jacobi identity (and hence define a Poisson structure on C ∞ (M )) if and only if
 ∂ Π ντ ρν ∂ Π
τµ
ρτ ∂ Π
µν 
Π ρµ + Π + Π = 0, ∀µ, ν, τ , (2.2)
ρ ∂ xρ ∂ xρ ∂ xρ
where in the coordinate system {xµ } we have Π = µν
µ,ν Π ∂xµ ∧ ∂xν .

Example 4 shows that every symplectic manifold is naturally a Poisson manifold. Example 3 shows that the dual of a Lie
algebra g is naturally a Poisson manifold (which typically will not be symplectic).
 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2261

2.2. Supermanifolds

For an introduction to supermanifolds we refer the reader to [9]. We here collect basic definitions and notations that will
be used below.

Definition 3. A commutative superalgebra A = A0̄ ⊕ A1̄ is a Z/2Z-graded vector space together with an associative bilinear
operation · : A ⊗ A → A satisfying a · b = (−1)p(a)p(b) b · a, ∀a, b ∈ A, where p(a) is the degree of a. Such an algebra is unital
if there exists an element 1 ∈ A0 such that 1 · a = a · 1, ∀a ∈ A. We will assume all of our algebras to be unital.

Example 5. Let B be an associative commutative algebra over F. By A = B[θ 1 , . . . , θ n ] we denote the free commutative
superalgebra over B generated by {θ i } 
⊂ A1̄ . By extension of scalars we will consider this algebra as an F-algebra. As a vector
Fn , where Fn is the Grassman algebra in n-generators.

space, this is the tensor product B ⊗F

Let C ∞ be the sheaf of smooth functions on Rp . The space Rp|q is the topological space Rp endowed with the sheaf of
commutative superalgebras C ∞ [θ 1 , . . . , θ q ].

Definition 4. A supermanifold M of dimension p|q is a topological space |M | endowed with a sheaf OM of commutative
superalgebras (over R) which is locally isomorphic to Rp,q .

Morphisms of supermanifolds are defined to be morphisms of ringed spaces, namely, f : M → N is the data of a
continuous map |f | : |M | → |N | together with a morphism of sheaves of superalgebras from f ∗ ON to OM (here the pullback
f ∗ is defined in the standard way).

Example 6. Let |M | be a smooth p-dimensional manifold and let E be a rank q vector bundle over |M |. Let C ∞ be the sheaf
of smooth functions on |M |, the sheaf of sections of E is a locally free C ∞ -module of rank q. Consider the exterior product
 ∗ ∗
E , this is a sheaf of commutative superalgebras (graded by the exterior degree modulo 2) over |M | which is locally of
the form C ∞ [θ 1 , . . . , θ q ] for some local frame {θ i } of E ∗ . As such it defines a supermanifold M of dimension p|q denoted by
Π E.
Below we will need the case when |M | is the circle S 1 and E is its tangent bundle. The corresponding supermanifold is
denoted in this case S 1|1 . Letting θ be a global 1-form on S 1 , we see that the sheaf of functions on S 1|1 is simply C ∞ ⊕ C ∞ · θ ,
where C ∞ is the sheaf of smooth functions on S 1 .

Many kinds of spaces are defined as topological spaces endowed with a sheaf of algebras. Depending on which kinds of
algebras we consider, we obtain different categories: smooth manifolds, holomorphic manifolds, super manifolds, etc. We
may consider a refinement of the above construction by considering Z-graded algebras instead of Z/2Z-graded ones (with
the same supercommutativity constrain), these are usually called graded supermanifolds. Typical examples are as Example 6
above, where the grading is now the exterior degree (instead of considering it modulo 2). Since the local sections θ i are of
degree 1, it is natural to call this graded manifold E [1]. More generally, given a graded supermanifold (M , OM ) and a locally
free OM -module E, we may consider for each integer n > 0 a graded supermanifold E [n], as a topological space it is still |M |
and its sheaf of graded algebras is the algebra freely generated (over OM ) by local sections of E ∗ placed in degree n. In the
particular case when E = TM (resp. E = T ∗ M) is the tangent bundle (resp. cotangent bundle) of a smooth manifold M, we
denote this graded supermanifold as T [n]M (resp. T ∗ [n]M).
A word of caution is in order: even though any graded supermanifold gives an example of a supermanifold (by considering
the grading modulo 2), the morphisms are much more restrictive in the former case, as we would only consider Z-graded
maps between Z-graded algebras, instead of the corresponding Z/2Z-graded objects. These subtleties will not play any role
in this article.

3. N = 2 SUSY vertex algebra

In this section, we briefly review the definitions of vertex algebras and their N = 2 supersymmetric counterparts. The
number of supersymmetries introduced are in general arbitrary, but since we are mainly interested in the case of two
supersymmetries in this work, we choose to be concrete. For more details, the reader is referred to [10,11].
Given a vector space V , a field is defined as an End(V )-valued distribution in a formal parameter z:
 1
A(z ) = A(j) , where A(j) ∈ End(V ), (3.1)
j∈Z
z j+1

and, for all B ∈ V , A(z )B contains only finitely many negative powers of z.
A vertex algebra is a vector space V (the space of states), with a vector |0⟩ ∈ V (the vacuum), a map Y from a given state
A ∈ V to a field Y (A, z ) (called the state-field correspondence), and an endomorphism ∂ : V → V (the translation operator).
The field Y (A, z ) will also be denoted by A(z ).
2262 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

These structures must fulfill a set of axioms. The vacuum should be invariant under translations: ∂|0⟩ = 0. Acting with
∂ on a field should be the same as differentiation of the field with respect to the formal parameter z:
[∂, Y (A, z )] = ∂z Y (A, z ). (3.2)
We will use ∂ to denote both the endomorphism and ∂z , and it should be clear from the context what we mean by ∂ . The
field Y (A, z ) corresponding to a given state A creates the same state from the vacuum in the limit z → 0:
Y (A, z )|0⟩ |z =0 = A(−1) |0⟩ = A. (3.3)
The construction easily extends to the case when V is a super vector space. The state-field correspondence Y should respect
this grading, ∂ should be an even endomorphism, and the vacuum should be even.
From the endomorphisms A(j) of Y (A, z ) (called the Fourier modes), we can define the λ-bracket:
 λj
[Aλ B] = (A(j) B), (3.4)
j ≥0
j!

where λ is an even formal parameter. The λ-bracket can be viewed as a formal Fourier transformation of Y (A, z )B:

[ A λ B ] = Resz eλz Y (A, z )B, (3.5)

−1
where Resz picks the z -part of the expression. The locality axiom of the vertex algebra says that the sum (3.4) is finite for
all A and B, in other words, all fields in a vertex algebra are mutually local.
The λ-bracket captures the operator product expansion of the corresponding (chiral) fields in a two dimensional quantum
field theory. Taking the residue in (3.5) picks out the pole in z, which can be considered to be a formal δ -function. The
parameter λ then keeps track of how many derivatives act on the δ -function. In other words, (3.4) is equivalent, in the
familiar notation of OPEs, to
 A(j) B (w)
 
A(z ) · B(w) ∼ . (3.6)
j ≥0
(z − w)j

3.1. NK = 2 supersymmetric vertex algebra

A vertex algebra endowed with extra supersymmetries can conveniently be described by the formalism of SUSY vertex
algebras. By introducing two additional formal parameters, θ 1 and θ 2 , that are odd, and promoting the fields A(z ) to
superfields A(z , θ 1 , θ 2 ), we obtain the notion of NK = 2 SUSY vertex algebra of [11]. In the following, we will often drop
the subscript K .
We let Z = (z , θ 1 , θ 2 ) and consider N = 2 superfields of the form
 1 
A(Z ) = A(j|11) + θ 1 A(j|01) + θ 2 A(j|10) + θ 1 θ 2 A(j|00) ,

(3.7)
j∈Z
z j +1

where A(j|∗∗) ∈ End(V ) and, as before, for all B ∈ V , A(Z )B contains only finitely many negative powers of z. The state-
field correspondence Y (A, Z ) maps a state A, to a superfield A(Z ). We have two odd endomorphisms: D1 and D2 satisfying
[Di , Dj ] = δij ∂ and [Di , ∂] = 0. The vacuum is translation invariant: Di |0⟩ = 0. We require translation invariance,
∂
 
[Di , Y (A, Z )] = − θ i
∂ z Y (A, Z ). (3.8)
∂θ i
In addition to the even formal parameter λ, we introduce two odd formal parameters, χ1 and χ1 , with the relations
[χi , χj ] = −2δij λ and [χi , λ] = 0. We can then define the N = 2 SUSY Λ-bracket:

[ A Λ B ] = resZ e(z λ+θ χ1 +θ χ2 ) Y (A, Z )B

1 2

 λj 
A(j|00) − χ1 A(j|10) + χ2 A(j|01) − χ1 χ2 A(j|11) B,

= (3.9)
j≥0
j!

where resZ is the coefficient of θ 1 θ 2 z −1 . The locality axiom of the SUSY vertex algebra requires that the sum (3.9) is finite
for all A and B, i.e., all fields in a SUSY vertex algebra are mutually local.
Let us define the normal ordered product : : between two states by
V ⊗ V → V, A ⊗ B →: AB :≡ A(−1|11) B. (3.10)
In the following, we will often omit the symbol : :, and use parenthesis to indicate when the ordering is important. Properties
of the normal ordering product and the relations between the Λ-bracket and the normal ordering are given in Appendix A.
 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2263

We note however that the normal ordered product is not associative nor commutative. The Λ-bracket and the normally
ordered product satisfy a Leibniz-like rule (A.7) known as the non-commutative Wick formula. In fact, one can define an
NK = 2 SUSY vertex algebra as a tuple (V , |0⟩, : : , [Λ ], D1 , D2 , ∂) satisfying the axioms of Appendix A.
If one drops the integral terms in the axioms, one arrives to the notion of a Poisson N = 2 SUSY vertex algebra
[11, Section 4.10]. In this case, one writes {Λ } for the Λ-bracket, and we note that V with its operation · becomes a unital
super-commutative associative algebra since the quantum corrections in (A.5) and (A.6) vanish. Moreover, the Poisson
Λ-bracket {Λ } now is distributive with respect to : : (i.e. the Leibniz rule holds) since the quantum correction in (A.7)
vanishes.
Let us consider the situation when one has a family Vh̄ of N = 2 SUSY vertex algebras parametrized by h̄, that is, an N = 2
SUSY vertex algebra over C[[h̄]], such that the fiber at h̄ = 0 is a Poisson vertex algebra V0 with the operations defined as
1
: AB ::= lim : AB :h̄ , {AΛ B} := lim [AΛ B]h̄ .
h̄→0 h̄→0 h̄
We then say that the family is a quantization of V0 , or that V0 is the quasiclassical limit of Vh̄ . This happens for example when
V is the universal enveloping SUSY vertex algebra of a conformal Lie algebra, namely, when V is generated by some fields
Ai such that their OPE only involves the fields Ai and their derivatives. In this situation, one may consider the algebra Vh̄
generated by the same {Ai } with the Λ-bracket

[Ai Λ Aj ]h̄ := h̄[Ai Λ Aj ].

We easily see that the quantum corrections of (A.5) and (A.6) are of order h̄, and therefore they vanish on V0 . We thus obtain
a quasiclassical limit of Vh̄ .

3.2. Example: the λ-brackets of an N = 2 superconformal vertex algebra

The N = 2 superconformal vertex algebra is generated by a Virasoro field L, two odd fields G+ and G− , an even field J,
and a central element c (the central charge) [10], with

λ3
 
3
[ L λ L ] = (∂ + 2λ)L + c, [ L λ Gi ] = ∂ + λ Gi , (3.11)
12 2
[ L λ J ] = (∂ + λ)J , (3.12)
λ2
 
1
[ G+ λ G− ] = L + λ + ∂ J + c , [ G± λ G± ] = 0, (3.13)
2 6
λ
[ J λ G± ] = ±G± , [J λ J ] = c. (3.14)
3
In an NK = 2 SUSY vertex algebra, the same algebra is generated by a single field G, with the Λ-bracket [11]
c
[ G Λ G ] = (2λ + 2∂ + χ1 D1 + χ2 D2 ) G + λχ1 χ2 , (3.15)
3
where the superfield G(Z ) is expanded as

G(Z ) = −iJ (z ) + iθ 1 G+ (z ) − G− (z ) − θ 2 G+ (z ) + G− (z ) + 2θ 1 θ 2 L(z ).

   
(3.16)

3.3. Graded SUSY vertex algebras

In this section, we recall the concepts of gradings by conformal weights and charge in the supersymmetric case. As always,
we restrict to the case of 2 supersymmetries, and we will omit the terms NK = 2 below.
Recall from [11, Definition 5.6] that a SUSY vertex algebra V is called conformal if it admits a vector τ ∈ V such that
defining G(Z ) = Y (τ , Z ) this field satisfies (3.15), and moreover
• τ(0|00) = 2∂, τ(0|10) = −D1 , τ(0|01) = D2 .
• The operator H := 21 τ(1|00) acts diagonally with eigenvalues bounded below and with finite dimensional eigenspaces.
In this case the eigenvalues of H are called the conformal weights. Moreover, it follows from [11, Theorem 4.16(4)] that,
∀a ∈ V ,
 
1 1
[H , Y (a, Z )] = z ∂z + θ ∂θ 1 + θ 2 ∂θ 2 Y (a, Z ) + Y (Ha, Z ).

(3.17)
2
2264 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

A SUSY vertex algebra will be called graded if there exists a diagonal operator H satisfying (3.17). If Ha = 1a for ∆ ∈ C
we say that a has conformal weight, or dimension, ∆. It is easy to see that in this case:
1
∆(∂ a) = ∆(a) + 1, ∆(Di a) = ∆(a) + , ∆(: ab :) = ∆(a) + ∆(b), (3.18)
2
and if we let ∆(λ) = 1 and ∆(χ i ) = 1/2 then all the terms of the Λ bracket [aΛ b] have conformal weight ∆(a) + ∆(b), so
that the OPE (or the Λ bracket) becomes a graded operation of degree zero. This is a special property of the N = 2 case, in
general the OPE is of degree N /2 − 1.
We want to construct a supersymmetric theory where the scalar fields consist of functions on the target manifold.
If we want these fields to have dimension zero, then it is clear that their OPE will vanish unless our theory is N = 2
supersymmetric. In this case, the Λ-bracket has to be another field of conformal weight zero. In particular, the Λ-bracket of
functions is an operation on functions.
In fact, the following is a simple exercise in SUSY vertex algebras:

Theorem 1. Let V be a graded NK = 2 SUSY vertex algebra such that the conformal weights are bounded by 0. Let V0 be the space
of conformal weight 0 vectors, then V0 is naturally a Poisson algebra, with multiplication being the normally ordered product, and
the Poisson bracket being the Λ-bracket.
Immediately we see that if we want the dimension zero sector of our theory to consist of functions on the target manifold
M then M has to be a Poisson manifold. This is the content of the next section.
The theorem above can be generalized as in the non-SUSY case. Indeed, given a SUSY vertex algebra V , it is easy to see
that
V
P (V ) := , (3.19)
: VD1 V : + : VD2 V :
is naturally a Poisson algebra, the associative commutative product is induced from the normally ordered product and the
Poisson bracket is induced from the (0|00)-th product. If V is graded, then P (V ) inherits the grading, and therefore the
zero-th weight space is a Poisson subalgebra.

4. Sheaf of N = 2 VA from a Poisson structure

In this section, we construct a sheaf of SUSY vertex algebras on any Poisson manifold (M , Π ). The heuristic is simple;
we first attach a local model to an affine space and then we need to prescribe how these local fields change under the
allowed local automorphisms (depending on whether we work in the algebraic, real-analytic or smooth setting). In [1],
the authors attach to Rn (in the smooth setting) and coordinates {xν } a free βγ − bc-system. That is a vertex algebra
generated by 2n fermionic fields {bν , c ν }, and 2n bosonic fields {γ ν , βν }. What the authors noticed is that under changes
of coordinates, the fields γ ν transform as the coordinates {xν } do, the fields bν (respectively c ν ) transform as the vector
fields ∂/∂xν do (respectively the differential forms dxν ). The fields βν , however, do not transform as tensorial objects, but in
a rather complicated way. In fact, one may think of the generating fields γ µ , βµ , c µ and bµ as coordinates on the graded
supermanifold M̃ := T ∗ [2]T [1]M and CDR may be thought of as a formal quantization of loops into this manifold.
It was noticed in [2] that if we instead of looking at 4n-fields as generators, we study 2n-superfields as generators, these
objects transform as tensors. This corresponds to trading supersymmetry in the target by supersymmetry in the worldsheet,
namely, instead of loops into M̃ as above, we are looking at N = 1 superloops (maps from S 1|1 ) into the supermanifold
M ′ := T ∗ [1]M. In terms of the previous generators (in the non-SUSY case), the superfields are given by
φν = γ ν + θ cν , Sν = bν + θ βν ,
where the superfields φ ν are even and transform as the coordinates {xν } do, while the superfields Sν are odd and transform
as the vector fields ∂/∂xν do.
In this article, we exploit further this mechanism by which we trade the complexity of each generator (they are superfields
with more components), by simplicity of the transformation formula under changes of coordinates. For this we will look at
N = 2 superloops into M. Locally, to Rn we will attach a SUSY vertex algebra generated by n superfields (N = 2) Φ ν (which
in components account for the 4n generators in the classical sense) such that they transform as coordinates do. It follows
from Theorem 1 that the OPE of these fields has to be of the form:
[ Φ µ Λ Φ ν ] = Π (Φ )µν , (4.1)
where Π µν are the components of a Poisson bivector. In fact, we have the following.

Theorem 2. Let M be a Poisson manifold and let O be its sheaf of smooth functions. There exists a sheaf V of SUSY vertex algebras
on M generated by O , together with an embedding ι : O → V , such that
ι(fg ) =: ι(f )ι(g ) :, ι{f , g } = [ι(f )Λ ι(g )], (4.2)
for all local sections f , g of O . This sheaf satisfies a universal property such that for any other sheaf V satisfying (4.2), then there
′

exists a unique surjective morphism j : V → V ′ .

 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2265

Proof. The proof of this statement is straightforward just as in the construction of the chiral de Rham complex [1]
(see also [3, Proposition 4.6]). Since the construction in the N = 2 supersymmetric case is simpler than in the non-SUSY
case of [1] and the N = 1 case of [3] we sketch here the proof. Locally, one can proceed as follows. For a Poisson algebra
O we consider the free H -module generated by O (see Appendix A for notation). This module has a structure of SUSY Lie
conformal algebra with the operation
[fΛ g ] := {f , g }, (4.3)
′
extended by sesquilinearity. We can consider its universal enveloping SUSY vertex algebra V [11]. We now consider its
quotient V by the ideal generated by the relations
fg =: fg :, Di (fg ) :=: (Di f )g : + : fDi g :, 1O = |0⟩, (4.4)
∀f , g ∈ O , i = 1, 2. Since the operations are defined locally, this ideal is compatible with localization and in fact we obtain
a sheaf locally described by this quotient V .
Notice that V ′ is naturally graded (declaring O to be of degree zero). Since the ideal (4.4) is homogeneous it follows that
V is also graded. In fact, we see that locally O is just the degree zero part of V . 

Remark 1. There is a subtlety when we say that this sheaf is generated by n-superfields satisfying (4.1). If we are in the
algebraic setting and the bivector Π is algebraic then we can use arguments of formal geometry to make sense of the RHS
of (4.1). In the smooth setting we may construct the sheaf as in the proof of the Theorem, or argue as in [12].
This sheaf of NK = 2 SUSY vertex algebras can also be viewed as a sheaf of vertex algebras, generated by the components
of Φ . Naming the components of Φ as

Φ µ = γ µ + θ 1 c µ + θ 2 dµ + θ 1 θ 2 δ µ , (4.5)
the bracket (4.1) is equivalent to the λ-brackets
[ γ µ λ δ ν ] = Π µν , [ c µ λ dν ] = Π µν , (4.6)
[ c µ λ δ ν ] = Π,τµν c τ , [ dµ λ δ ν ] = Π,τµν dτ , (4.7)
1
[ δ µ λ δ ν ] = Π,τµνρ (dτ c ρ − c τ dρ ), (4.8)
2
where Π is evaluated at γ and the rest of the brackets are zero. Note that, for a linear Poisson-structure the δ ’s commute.
Here γ is even, and transforms as a coordinate. The odd fields c and d transform as vectors, and the even field δ transforms
in an inhomogeneous way.
Alternatively, we can generate the sheaf by N = 1 superfields. Expand Φ as Φ µ = φ µ (z , θ 1 ) − θ 2 S µ (z , θ 1 ). We then
have
[ φ µ Λ S ν ]NK =1 = Π µν , [ S µ Λ S ν ]NK =1 = Π,τµν S τ . (4.9)
This shows that the Poisson calculus, in the sense of [13], can be mapped to the NK = 1 vertex algebra corresponding to
(4.1). For any Poisson manifold, the cotangent bundle is equipped with the non-trivial structure of Lie algebroid. Namely, in
local coordinates we have
{dxµ , dxν } = Π,τµν dxτ , {f (x), dxµ } = Π µν ∂ν f , (4.10)
where f (x) ∈ C ∞ (M ) and dx is the local basis for differential forms. Thus, on a Poisson manifold one can construct
the Courant algebroid (bi-algebroid TM ⊕ T ∗ M with the above bracket on TM and the trivial bracket on T ∗ M) and the
corresponding sheaf of N = 1 SUSY vertex algebras is generated by the relations (4.9).

4.1. Relation to the Chiral de Rham complex

The N = 2 sheaf can be related to the Chiral de Rham complex (the sheaf of N = 1 SUSY vertex algebras associated to
the standard Courant algebroid on TM ⊕ T ∗ M). It is instructive to expand the superfield Φ in such way so we make contact
with previous [3,2,7] calculations.
Let φ µ (z , θ 1 ) be an even N = 1 superfield, and Sν (z , θ 1 ) an odd N = 1 superfield with the expansions

φ µ (z , θ 1 ) = γ µ (z ) + θ 1 c µ (z ), (4.11)
and
Sµ (z , θ 1 ) = bµ (z ) + θ 1 βµ (z ). (4.12)
The field φ µ (z , θ 1 ) transforms as a coordinate, and Sν (z , θ 1 ) as a one-form. Recall that the defining Λ-bracket of the Chiral
de Rham complex is
[ φ µ Λ Sν ]NK =1 = δνµ , (4.13)
2266 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

with [ φ µ Λ φ ν ]NK =1 and [ Sµ Λ Sν ]NK =1 being zero. Written as λ-brackets, e.g., with no manifest supersymmetry, this is

[ βν λ γ µ ] = δνµ , [ c µ λ bν ] = δνµ , (4.14)

and the rest of the brackets are zero.
From these brackets and fields, we can construct an N = 2 superfield Φ µ , that will fulfill (4.1), by

Φ µ (z , θ 1 , θ 2 ) = φ µ (z , θ 1 ) − θ 2 Π µν (φ(z , θ 1 ))Sν (z , θ 1 ). (4.15)

In components, this is

Φ µ = γ µ + θ 1 c µ − θ 2 Π µν bν + θ 1 θ 2 (Π µν βν + (Π,τµν c τ )bν ). (4.16)

If the Poisson structure is degenerate, this Φ may differ from the most general Φ fulfilling (4.1), and it is only on a symplectic
manifold where the sheaf generated by (4.1) is the same as the CDR.

4.2. Automorphism of the algebra

The labeling of the two θ ’s in the definition of the SUSY vertex algebra is arbitrary, and when we have more than one
supersymmetry, we also have an R-symmetry. In particular, the bracket (4.1) is invariant under the transformations

θ 1 → −θ 2 , θ 2 → θ 1. (4.17)
If we also let D1 → −D2 and D2 → D1 , then axiom (3.8) is still satisfied. This automorphism may induce non-trivial
transformations on the components of the superfields.

4.3. Quasi-classical limit

The sheaf V constructed above admits a quasi-classical limit P as a sheaf of SUSY Poisson vertex algebras. It is generated
by O just as in (4.2) with the normally ordered product : : replaced by the associative commutative product of the Poisson
vertex algebra and its Λ-bracket [Λ ] replaced by the Poisson Λ-bracket {Λ }.

5. N = 2 algebra on a symplectic manifold

In this section, we discuss the case of a symplectic structure. If the Poisson bivector Π is invertible, then M is symplectic
and we will use a different notation for this case: Π µν = ωµν . The symplectic structure ωµν is a closed non-degenerate two
form, such that ωµν ωνρ = δρµ . We can then associate a sheaf of N = 2 vertex algebras to the manifold, generated by

[ Φ µ Λ Φ ν ] = ω(Φ )µν . (5.1)

The symplectic case is interesting since we have a canonical two form ωµν . From the Φ ’s, we can construct objects that
transform as vectors, Di Φ µ , or ∂ Φ µ . To construct target space diffeomorphism invariant operators, currents, out of these
objects, we need tensors with covariant indices that we can contract with, e.g., forms. The most apparent example to study
is the case of a symplectic manifold.
As noted above, this sheaf is essentially the same as the Chiral de Rham complex. If we expand Φ as in (4.15), we can use
ω to project out Sν . The brackets (4.13) and (5.1) are then equivalent.
The automorphism (4.17) induces an automorphism on the components of Φ , given by
γ µ → γ µ, βµ → βµ + (ωτ ν,µ ωνσ )(c τ bσ ) + ωµσ ,ν ∂ωνσ , (5.2)
µ µν ν
c → −ω bν , bµ → ωµν c . (5.3)
This automorphism of the βγ − bc-system was discovered, in the case of a Calabi–Yau target manifold, in [3, Theorem 6.4].

5.1. N = 2 superconformal algebra

On the symplectic manifold, the sheaf carries the structure of an N = 2 superconformal algebra. We can construct a
generator Gω by
1
Gω = ωµν (D1 Φ µ D1 Φ ν + D2 Φ µ D2 Φ ν ) . (5.4)
2
There are no order ambiguities in this expression. The operator Gω is a well defined section of the sheaf, and there is no need
for any quantum corrections. The operator fulfill the N = 2 superconformal algebra
c
[ Gω Λ Gω ] = (2λ + 2∂ + χ1 D1 + χ2 D2 ) Gω + λχ1 χ2 , (5.5)
3
with central charge c = 3 dim M. The proof is given in Appendix B.
 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2267

6. N = (2, 2) vertex algebra on a Calabi–Yau manifold

Let us consider a Kähler manifold M, with Kähler form ω. Consider the NK = 2 SUSY vertex algebra generated by

[ Φ α Λ Φ β̄ ] = ωαβ̄ . (6.1)
α β̄
Here the fields Φ and Φ correspond to holomorphic and anti-holomorphic coordinates. Let us define an operator H0 by

H0 = (gα β̄ D2 Φ α )D1 Φ β̄ − (gαβ̄ D1 Φ α )D2 Φ β̄ . (6.2)

As it stands, this operator is not a well-defined section of the sheaf of vertex algebras for a general Kähler manifold. It may
need a ‘‘quantum correction’’, as we will see soon. At this stage, the operator might seem rather ad-hoc, but we will motivate
it by the discussion of sigma model in Section 7.
In order to construct a well defined section of the sheaf of vertex algebras, we need to investigate how H0 transforms
under coordinate changes. Let {z α } be a holomorphic coordinate system, and let z̃ α = F α (z β ) be an invertible, holomorphic
α β̄
change of coordinates. We have g̃δ ε̄ = gα β̄ Φ,δ Φ,ε̄ and

β̄ β̄
α
g̃δ ε̄ D2 Φ̃ δ = (gα β̄ Φ,δ δ
Φ,ε̄ )(Φ̃,γ D2 Φ γ ) = gα β̄ Φ,ε̄ D2 Φ α , (6.3)
and, using quasi-associativity (A.6),
β̄
 
g̃δ ε̄ D2 Φ̃ δ D1 Φ̃ ε̄ = gα β̄ Φ,ε̄ D2 Φ α Φ̃,ε̄ρ̄ D1 Φ ρ̄
  

 ∇ 
β̄ β̄
  
= gα β̄ Φ,ε̄ D2 Φ α Φ̃,ε̄ρ̄ D1 Φ ρ̄ − dΛΦ̃,ε̄ρ̄ [ gαβ̄ Φ,ε̄ D2 Φ α Λ D1 Φ ρ̄ ]
0

ρ̄
 
= gαβ̄ D2 Φ α D1 Φ β̄ − i ∂ Φ̃,ε̄ρ̄ Φ,ε̄ .
 
(6.4)

Therefore, under the inverse change of coordinates z̃ → z , H0 transforms as

 
ρ̄ ∂ det Ā
H0 → H0 − 2i ∂ Φ̃,ᾱρ̄ Φ,ᾱ = H0 − 2i , (6.5)
det Ā
where Aα β = ∂ z̃ α /∂ z β is the Jacobian of the change of coordinates and Ā is its complex conjugate. We see immediately that
H0 will define a global section of our sheaf if M is Calabi–Yau. In that case, this section looks like (6.2) in the holomorphic
coordinate system where the holomorphic volume form is constant.
To find the expression for this section in a general holomorphic coordinate system we must add a quantum correction to
H0 that cancels the inhomogeneous transformations. On a Calabi–Yau manifold, we can write the volume form as Ω ∧ Ω̄ ,
where Ω is a holomorphic volume form, Ω = ef (z ) dz 1 ∧ · · · ∧ dz d/2 . Under the change of coordinates z̃ → z , f transforms
as a density:
α
f˜ = f + log det Φ,β = f − log det A. (6.6)
We can use this to cancel the inhomogeneous transformation of H0 . Thus, in general holomorphic coordinates of a
Calabi–Yau manifold,

H = H0 − 2i∂ f¯ = (gα β̄ D2 φ α )D1 φ β̄ − (gα β̄ D1 φ α )D2 φ β̄ − 2i∂ f¯ (6.7)

form a well defined section.
Let us now define G± by

1
G± = Gω ∓ H, (6.8)
2
where Gω is the operator constructed in (5.4). Introducing new odd derivatives, D± , that are linear combinations of the
derivatives D1 and D2 , by

1
D± ≡ √ (D1 ∓ iD2 ), (6.9)
2
we can write (6.8) in a general holomorphic coordinate system as

G± = (ωα β̄ D± Φ α )D∓ Φ β̄ ± i∂ f¯ . (6.10)

The following is the main result of [3] now stated in manifest N = 2 formalism. The proof can be found in Appendix C.
2268 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

Theorem 3. Let M be a Calabi–Yau manifold and G± be defined by (6.10). The sections G± generate two commuting N = 2
superconformal algebras,
c
[ G± Λ G± ] = (2λ + 2∂ + χ1 D1 + χ2 D2 ) G± + λχ1 χ2 ,
3 (6.11)
[ G± Λ G∓ ] = 0,
3
each with a central charge c = 2
dim M.

7. The N = 2 Hamiltonian of an N = (2, 2) supersymmetric sigma model

We now want to relate the above discussion to the Hamiltonian treatment of the supersymmetric sigma model, and
thereby motivate the expression (6.2). To do this, we consider the classical supersymmetric sigma model, and we derive a
Hamiltonian formulation thereof. A similar treatment of the N = 1 sigma model was initiated in [14,15] and its relation to
CDR was suggested in [7]. Here, we suggest the similar relation between the N = (2, 2) supersymmetric sigma models with
a Calabi–Yau target and the sheaf of N = 2 supersymmetric vertex algebras on the same Calabi–Yau.
We restrict ourself to the N = (2, 2) supersymmetric sigma model with the target manifold M being a Kähler manifold,
which is not the most general sigma model with this amount of supersymmetry. The action functional for a classical
N = (2, 2) supersymmetric sigma model is given by

S= dσ dτ dθ+
1
dθ−
1
dθ+
2
dθ−
2
K (Φ , Φ̄ ), (7.1)

where the integral performed over Σ 2|4 with even coordinates t , σ and four odd θ coordinates. For the sake of simplicity,
we assume that Σ = R × S 1 . Φ and Φ̄ are maps from Σ 2|4 to M which satisfy some first order differential equation
(see Appendix D). In physics, Φ = {Φ α } is called a chiral superfield, and Φ̄ = {Φ ᾱ } is an anti-chiral superfield. K is the
Kähler potential, which is defined only locally, but nevertheless the action functional (7.1) is well-defined. Upon integration
of the odd θ -coordinates, the functional (7.1) reduces to the more familiar form of the non-linear sigma model and its
critical points are the generalizations of harmonic maps from Σ to M. In Appendix D, we set the notation and present some
properties of this N = (2, 2) model which are needed for the derivation. For more on supersymmetric sigma models and
their applications, the reader may consult the book [16].
We would like to consider the Hamiltonian formulation of (7.1). By doing a change of the odd variables, and integrating
out two of them, the action (7.1) can be written as
  
α 1
S= dσ dτ dθ dθ2 1
iK,α ∂0 φ − H , (7.2)
2
with
H = gαβ̄ D2 φ α D1 φ β̄ − gαβ̄ D1 φ α D2 φ β̄ (7.3)
being the Hamiltonian and ∂0 being the derivative along τ (time). Here, θ 1 and θ 2 , with corresponding odd derivatives,
Di = ∂θ∂ i + θ i ∂σ , are the remaining two odd coordinates. Also, K,α β̄ = gα β̄ . See Appendix D for a more detailed derivation.
From (7.2), we see that the Poisson bracket is given by

{φ α , φ β̄ } = ωαβ̄ , (7.4)
and that the Hamiltonian density of the sigma model is given by (7.3). The bracket (7.4) is the same as the bracket of the
Poisson vertex algebra corresponding to the vertex algebra generated by (6.1). The expression (7.3) is the classical version
of the operator H considered in (6.7). Thus, following the logic presented in [7], we can think of the sheaf of N = 2
supersymmetric vertex algebras on a Calabi–Yau as a formal quantization of the N = (2, 2) sigma model defined by the
action (7.1).

8. Summary and discussion

In this note, we construct a sheaf of N = 2 supersymmetric vertex algebras for a Poisson manifold. We also study the
properties of this sheaf on symplectic and Calabi–Yau manifolds. We relate the corresponding semiclassical limit to the
N = (2, 2) non-linear sigma model. Let us conclude with a few remarks.
• As mentioned above, given an NK = 2 SUSY vertex algebra V , the quotient P (V ) defined by (3.19) is a Poisson algebra. Just
as in the non-SUSY case, there exists an analogous construction of the Zhu algebra associated to V , this is a one parameter
family of associative superalgebras Ph̄ (V ) such that the special fiber h̄ = 0 coincides with P (V ) and all other fibers are
isomorphic. In general it is not true that this family is flat, or that Ph̄ (V ) is a deformation of the Poisson algebra P0 (V ).
However, given the construction in this article, starting from a Poisson manifold M with its sheaf of Poisson algebras O ,
we constructed a sheaf of SUSY vertex algebras V and we obtain a one parameter family of associative algebras Ph̄ (V ).
We easily see that P0 (V ) = O .
 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2269

This immediately leads one to question whether this family is indeed a deformation in this particular case, obtaining thus
a natural way of quantizing Poisson manifolds. We plan to return to this topic in a future publication.
• The most general N = (2, 2) non-linear sigma models are related to generalized Kähler geometry [17]. Thus, there should
be an analogous Hamiltonian treatment of these general models, and it should suggest how to define sheaves of N = 2
Poisson vertex algebras for a wider class of manifolds. However, it may require a bigger set of fields than considered in
this article. This problem remains to be studied.

Acknowledgments

M.Z. thanks KITP, Santa Barbara where part of this work was carried out. The research of M.Z. is supported by VR-grant
621-2008-4273 and is supported in part by DARPA under Grant No. HR0011-09-1-0015 and by the National Science
Foundation under Grant No. PHY05-51164.

Appendix A. Rules for Λ-brackets in NK = 2 SUSY vertex algebras

In this appendix, we collect some properties of Λ-bracket calculus. For further explanations and details, the reader may
consult [11].

• The operators Di , ∂ and the parameters χj , λ, where i, j = 1, 2, have the commutator relations [∂, χi ] = [Di , λ] =
[∂, λ] = 0, and
[Di , Dj ] = 2δij ∂, [χi , χj ] = −2δij λ, [Di , χj ] = 2δij λ. (A.1)

We will denote by H the super-algebra with two odd generators D1 , D2 and one even generator ∂ = [D1 , D2 ] commuting
with both D1 and D2 .
• Sesquilinearity:
[ Di a Λ b ] = −χi [ a Λ b ], [ a Λ Di b ] = (−1)a (Di + χi ) [ a Λ b ], (A.2a)
[ ∂ a Λ b ] = −λ[ a Λ b ], [ a Λ ∂ b ] = (∂ + λ) [ a Λ b ]. (A.2b)

• Skew-symmetry:

[ a Λ b ] = −(−1)ab [ b −Λ−∇ a ]. (A.3)

The bracket on the right hand side is computed as follows: first compute [ b Γ a ], where Γ = (γ , η), then replace Γ by
(−λ − ∂, −χ − D).
• Jacobi identity:

[ a Λ [ b Γ c ] ] = [ [ a Λ b ] Γ +Λ c ] + (−1)ab [ b Γ [ a Λ c ] ] (A.4)

where the first bracket on the right hand side is computed as in (A.3).
An H -module with an operation [Λ ] satisfying sesquilinearity, skew-symmetry, and the Jacobi identity is called a SUSY
Lie conformal algebra.
• Quasi-commutativity:
 0
ab − (−1)ab ba = [ a Λ b ]dΛ, (A.5)
−∇

where the integral −∇ dΛ is defined as ∂χ∂ ∂χ∂ −∂ dλ.

0 0
1 2
• Quasi-associativity:
 ∇   ∇ 
(ab)c − a(bc ) = dΛa [ b Λ c ] + (−1)ab dΛb [ a Λ c ]. (A.6)
0 0

• Quasi-Leibniz (non-commutative Wick formula):

 Λ
[ a Λ bc ] = [ a Λ b ]c + (−1)ab b[ a Λ c ] + [ [ a Λ b ] Γ c ]dΓ . (A.7)
0
2270 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

Appendix B. N = 2 algebra on a symplectic manifold

We want to show that

1
Gω = ωµν (D1 Φ µ D1 Φ ν + D2 Φ µ D2 Φ ν ) (B.1)
2
fulfill
[ Gω Λ Gω ] = (2λ + 2∂ + χ1 D1 + χ2 D2 ) G + λχ1 χ2 dim M , (B.2)
using the bracket

[ Φ µ Λ Φ ν ] = ω(Φ )µν , (B.3)

µν
where ω ωντ = δτµ .
Note that there are no ambiguities in the order of the normal ordering in (B.1). Since each term only
contains one type of D, there can be no χ1 χ2 -terms when the brackets of the constituents are calculated. Thus, no terms
survive the integration in (A.6).
We are free to choose any coordinates we want. Since we are on a symplectic manifold, we can choose Darboux
coordinates, where ω is constant. This simplifies the calculations considerably. Let

1
Gi ≡ ωµν Di Φ µ Di Φ ν . (B.4)
2
We first want to calculate [ Gi Λ Gi ]. We have [ Di Φ µ Λ Di Φ ν ] = λ ωµν , so [ Di Φ µ Λ Gi ] = λ Di Φ µ . Skew-symmetry then
gives

[ Gi Λ Di Φ µ ] = (λ + ∂) Di Φ µ . (B.5)
From this, we see that
[ Gi Λ Gi ] = (2λ + ∂) Gi . (B.6)
µ µ
We now want to calculate [ G1 Λ G2 ]. We have [ D2 Φ Λ G1 ] = −χ2 χ1 D1 Φ . Using skew-symmetry, we then get
[ G1 Λ D2 Φ ] = (∂ + χ2 D2 + χ1 D1 ) D2 Φ + χ2 χ1 D1 Φ µ .
µ µ
(B.7)
From this we see that

µ ν µ ν µ ν µ ν
[ G1 Λ D2 Φ D2 Φ ] = (∂ + χ2 D2 + χ1 D1 ) (D2 Φ D2 Φ ) + χ2 χ1 (D1 Φ D2 Φ + D2 Φ D1 Φ ) + , (B.8)

where the integral term is given by

 Λ  Λ
µ µ ν
[ (∂ + χ2 D2 + χ1 D1 ) D2 Φ + χ2 χ1 D1 Φ Γ D2 Φ ]dΓ = − χ1 χ2 [ D1 Φ µ Γ D2 Φ ν ]dΓ
0 0

= −λχ1 χ2 ωµν . (B.9)

From (B.8), it is now easy to see that

dim M
[ G1 Λ G2 ] = (∂ + χ2 D2 + χ1 D1 ) G2 + χ2 χ1 ωµν D1 Φ µ D2 Φ ν + λχ1 χ2 , (B.10)
2
and, finally,
[ Gω Λ Gω ] = [ G1 Λ G1 ] + [ G2 Λ G2 ] + [ G1 Λ G2 ] + [ G2 Λ G1 ]
= (2λ + 2∂ + χ1 D1 + χ2 D2 ) G + λχ1 χ2 dim M . (B.11)

Appendix C. N = (2, 2) algebra on a Calabi–Yau manifold

We want to calculate the algebra generated by G+ and G− , defined in (6.10), under the bracket (6.1). We are free to
work in any coordinates we want. A convenient choice is to choose the coordinates where the holomorphic volume form
is constant. On a Calabi–Yau, we can always choose such coordinates locally. In this coordinates, the quantum correction
α
±i∂ f¯ (z ) vanishes. Also note that Γαβ = 0 in these coordinates. To the metric, we have a corresponding Kähler potential K .
Let subscripts of K denote derivatives: Kµ1 ···µk ≡ ∂µ1 · · · ∂µk K , so Kα β̄ = gα β̄ = iωα β̄ , with g being the metric and ω the
Kähler form of the manifold.
Let us define pα ≡ iKα , and

G0± = D∓ pα D± φ α , M = iKαβ D+ φ α D− φ β . (C.1)

 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2271

We then have

G± = G0± ± M. (C.2)

Note that M vanishes for a flat manifold. The definition of p implies the brackets

[ φ α Λ pβ ] = δβα , [ φ ᾱ Λ pβ ] = iωᾱα Kαβ , (C.3a)

in addition to

[ φ α Λ φ β̄ ] = ωαβ̄ . (C.3b)

In light of the derivation of the Hamiltonian density in Section 7, pα can be understood as the conjugate momenta to Φ α ,
and the brackets (C.3) are the corresponding Dirac brackets; see (D.25).
Let us define linear combinations of χ1 and χ2 , to better suit the base (6.9):

1
χ± = √ (χ1 ± iχ2 ). (C.4)
2
The relations between D± and χ± are

[D± , D∓ ] = 2∂, [χ± , χ∓ ] = −2λ, [D± , χ± ] = 2λ, (C.5a)

[D± , D± ] = 0, [χ± , χ± ] = 0, [D± , χ∓ ] = 0. (C.5b)

Note that the rules of sesquilinearity give

[ D± a Λ b ] = −χ∓ [ a Λ b ], [ a Λ D± b ] = (−1)a (D± + χ∓ ) [ a Λ b ]. (C.6)

We want to prove that G+ and G− give two commuting N = 2 superconformal algebras, i.e.

d
[ G± Λ G± ] = (2λ + 2∂ + χ+ D+ + χ− D− ) G± + λχ1 χ2 ,
2 (C.7)
[ G± Λ G∓ ] = 0 .
We first prove that G0+ and G0− fulfill the algebra (3.15). In terms of the split (C.2), we then need to prove that

[ G0± Λ M ] + [ M Λ G0± ] ± [ M Λ M ] = (2λ + 2∂ + χ+ D+ + χ− D− ) M, (C.8a)

[ G∓ Λ M ] − [ M Λ G± ] ∓ [ M Λ M ] = 0.
0 0
(C.8b)

C.1. Algebra of G0±

The calculation of [ G0± Λ G0± ] is straightforward. We do the calculation for G0+ , the calculation for G0− can be deduced by
exchanging + and −. We have

[ pα Λ G0+ ] = χ− D− pα , [ G0+ Λ pα ] = (χ− + D+ )D− pα , (C.9a)

α α α α
[φ Λ G+ ] = χ+ D+ φ ,
0
[ G+ Λ φ ] = (χ+ + D− )D+ φ ,
0
(C.9b)

and

(χ± + D∓ )(χ∓ + D± ) = −χ∓ χ± − D± D∓ + 2∂ + χ± D± − χ∓ D∓ , (C.10)

so, remembering that (D± ) = 0, 2


[ G0+ Λ G0+ ] = [ G0+ Λ D− pα ]D+ φ α + D− pα [ G0+ Λ D+ φ α ] +

= ((χ+ + D− )([ G0+ Λ pα ]))D+ φ α + D− pα ((χ− + D+ )[ G0+ Λ φ α ]) +

= −χ− χ+ G0+ + ((2∂ + χ+ D+ )D− pα )D+ φ α − χ+ χ− G0+ + D− pα ((2∂ + χ− D− )D+ φ α ) +

= (2λ + 2∂ + χ+ D+ + χ− D− )G0+ + . (C.11)
2272 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

The integral term is given by

 
α
[ [ G+ Λ D− pα ] Γ D+ φ ]dΓ =
0
[ (2∂ + χ+ D+ − χ− χ+ )D− pα Γ D+ φ α ]dΓ
 
d
= − (χ− χ+ + 2γ )η+ η− δαα dΓ = i (χ− χ+ + 2γ )η1 η2 dΓ
2
d d
= −iλ(χ− χ+ + λ) = λχ1 χ2 . (C.12)
2 2
To see that G0+ and G0− commute, we note that

(χ± + D∓ )2 = 0, (C.13)

so

[ G0+ Λ D+ pα ] = (χ− + D+ )[ G0+ Λ pα ] = (χ− + D+ )2 D− pα = 0, (C.14a)

α α α
[ G+ Λ D+ φ ] = (χ+ + D− )[ G+ Λ φ ] = (χ+ + D− ) D+ φ = 0.
0 0 2
(C.14b)

Thus,

[ G0+ Λ G0− ] = [ G0+ Λ D+ pα ]D− φ α + D+ pα [ G0+ Λ D− φ α ] = 0. (C.15)

There is no integral term.

C.2. Algebra of G0± and M

Let us define some shorthand notation, and calculate some brackets we are going to use later. Let

Bαβ ≡ D+ φ α D− φ β , (C.16)

so M can be written as

M = iKαβ Bαβ . (C.17)

Let
σ
E± ≡ Γαβ Kσ γ D± φ γ Bαβ . (C.18)

Also, note that

γ
[ Kαβ Λ φ γ ] = iΓαβ . (C.19)

C.2.1. The bracket [ M Λ M ]

We want to calculate [ M Λ M ]. Since both the first and second arguments of the bracket have the same expression, M ,
we only need to calculate the poles represented by an odd number of λ’s and χ ’s, and from skew-symmetry we can deduce
the full answer. We have

[ M Λ M ] = i[ M Λ Kαβ ]Bαβ + iKαβ [ M Λ Bαβ ] + , (C.20)

where represents the integral term in the quasi-Leibniz.
First term of (C.20). We start with the first term in (C.20), so we want to calculate [ M Λ Kαβ ]. Now,

[ Kαβ Λ M ] = i[ Kαβ Λ Kγ δ ]Bγ δ + iKγ δ [ Kαβ Λ Bγ δ ]. (C.21)

We then need to calculate [ Kαβ Λ Bγ δ ]:


[ Kαβ Λ Bγ δ ] = [ Kαβ Λ D+ φ γ ]D− φ δ + D+ φ γ [ Kαβ Λ D− φ δ ] +

γ
= i(D+ + χ− )(Γαβ )D− φ δ + iD+ φ γ (D− + χ+ )(Γαβ
δ
)+ . (C.22)
 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2273

The integral term of (C.22) is

 Λ  Λ
γ δ
[ [ Kαβ Λ D+ φ ] Γ D− φ ]dΓ = [ (D+ + χ− )[ Kαβ Λ φ γ ] Γ D− φ δ ]dΓ
0 0
 Λ
γ
= i(−η− η+ )[ Γαβ Γ φ δ ]dΓ
0
γ
= −λ[ Γαβ Γ φ δ ]. (C.23)
From (C.21), using skew-symmetry, we have
δ γ γ
[ M Λ Kαβ ] = χ+ Γαβ Kγ δ D+ φ γ − χ− Γαβ Kγ δ D− φ δ − λ iKγ δ [ Γαβ Γ φ δ ]
γ
− D+ φ γ D− Kγ δ Γαβ
δ
− D+ Kγ δ Γαβ D− φ δ + · · · , (C.24)
where the dots represent terms with no poles, or no odd derivatives, or containing only terms where D± hits holomorphic
φ . So, using the notation defined in (C.18), the first term in (C.20) is
γ
χ+ iE+ − χ− iE− + λ Kγ δ [ Γαβ Γ φ δ ]Bαβ + O (λ0 ). (C.25)

Second term of (C.20). To calculate the second term of (C.20), we first calculate [ Bαβ Λ M ] using (C.22):
αβ αβ γδ
[B Λ M ] = i[ B Λ Kγ δ ] B
β
= χ+ Γγαδ D+ φ β Bγ δ − χ− Γγ δ D− φ α Bγ δ − λ i[ Γγαδ Γ φ β ]Bγ δ + O (λ0 ). (C.26)
The second term of (C.20) then is
γ
χ+ iE+ − χ− iE− + λ Kγ δ [ Γαβ Γ φ δ ]Bαβ + O (λ0 ). (C.27)
Integral term of (C.20). There will be an integral term in (C.20), given by
 Λ
i [ [ M Λ Kαβ ] Γ Bαβ ]dΓ . (C.28)
0

Skew-symmetry still guarantees that we only need to calculate the poles represented by an odd number of λ’s and χ ’s.
The integral gives at least λ, and the possible poles then are λ and λχ1 χ2 . Higher poles are not possible due to dimensional
arguments. Let M K ≡ [ M Λ Kαβ ]. We have

[ MK Γ Bαβ ] = [ MK Γ D+ φ α ]D− φ β + D+ φ α [ MK Γ D− φ β ] + . (C.29)

The integral term cannot be relevant here, since this would give at least a γ , and the integration in (C.28) would give at least
λ2 , but the highest possible power of λ is one. Recall that the only terms surviving the integration are the η+ η− -terms.
We first calculate the first term in (C.29). We have
[ MK Γ D+ φ α ] = (η− + D+ )[ MK Γ φ α ]. (C.30)
α α
So, we need the η+ - and η+ η− -part of [ M K Γ φ ], which can be found by looking at the corresponding terms of [ φ Γ M K ].
These, in turn, can be found by using (C.24), and we get
[ φ α Γ [ M Λ Kαβ ] ]η+ ,η+ η− = η+ [ φ α Γ Kγ δ ]Γαβ
δ
D+ φ γ

= −iη+ Γγαδ Γαβ

δ
D+ φ γ , (C.31)
and
[ MK Γ D+ φ α ]η+ η− = −iη− η+ Γγαδ Γαβ
δ
D+ φ γ , (C.32)
so the relevant part of the first term of (C.29) is

− iη− η+ Γγαδ Γαβ

δ γβ
B . (C.33)
The second term can be calculated by exchanging + and −, and yields the same term. In total, the integral term is
 Λ
2η− η+ Γγαδ Γαβ
δ γβ
B dΓ = −2iλ Γγαδ Γαβ
δ γβ
B . (C.34)
0

In total. Summing the contributions, and using skew-symmetry, we have

[ M Λ M ] = A + χ+ 2iE+ − χ− 2iE− + λ 2Q
= −A + (χ+ + D− )2iE+ − (χ− + D+ )2iE− + (λ + ∂)Q , (C.35)
2274 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

where A is the part of the bracket with no λ’s or χ ’s, and

γ γ
Q ≡ (Kγ δ [ Γαβ Γ φ δ ] − iΓαδ Γγδβ )Bαβ . (C.36)
Thus
[ M Λ M ] = (2χ+ + D− )iE+ − (2χ− + D+ )iE− + (2λ + ∂)Q . (C.37)

C.2.2. The bracket [ M Λ G0+ ]

We want to calculate [ M Λ G0+ ] + [ G0+ Λ M ], and later [ M Λ G0− ] − [ G0+ Λ M ]. We start with [ M Λ G0+ ], and the other
terms can then be calculated by using skew-symmetry and by exchanging + and −. Using Leibniz and sesquilinearity, we
see that

γ γ
[ M Λ G+ ] = [ M Λ D− pγ ]D+ φ + D− pγ [ M Λ D+ φ ] +
0


= (χ+ + D− )([ M Λ pγ ])D+ φ γ + D− pγ (χ− + D+ )([ M Λ φ γ ]) + . (C.38)

First part of (C.38). We first note that

[ pγ Λ M ] = i[ pγ Λ Kαβ ]Bαβ + iKαβ [ pγ Λ Bαβ ], (C.39)

with no integral term, and

[ pγ Λ Bαβ ] = χ+ δγβ D+ φ α − χ− δγα D− φ β , (C.40)

so
[ pγ Λ M ] = i[ pγ Λ Kαβ ]Bαβ + iχ+ Kγ α D+ φ α − iχ− Kγ α D− φ α . (C.41)
Using skew-symmetry, we have

[ M Λ pγ ] = −i[ pγ Λ Kαβ ]Bαβ + i(χ+ + D− )(Kγ α D+ φ α ) − i(χ− + D+ )(Kγ α D− φ α ). (C.42)

From (C.5), we see that (χ± + D∓ ) = 0, and we note that 2

(χ+ + D− )(χ− + D+ ) = χ+ D+ − χ− D− − χ− χ+ + D− D+ , (C.43)

so, the first part of (C.38) is
− i(χ+ + D− )([ pγ Λ Kαβ ]Bαβ )D+ φ γ − i(χ+ D+ − χ− D− − χ− χ+ + D− D+ )(Kγ α D− φ α )D+ φ γ

= −i(χ+ + D− )([ pγ Λ Kαβ ]Bαβ )D+ φ γ + χ+ D+ M − χ− χ+ M − iχ− D− Kαβ Bαβ − iD− D+ (Kγ α D− φ α )D+ φ γ . (C.44)
Second part of (C.38). We have
γ
[ φγ Λ M ] = i[ φ γ Λ Kαβ ]Bαβ = Γαβ Bαβ , (C.45)
so, the second part of (C.38) is
γ
− D− pγ (χ− + D+ )(Γαβ Bαβ ). (C.46)
In the coordinates chosen, this is
γ γ
(χ− + D+ )(Γαβ Bαβ )D− pγ − i∂([ pγ Λ Γαβ ]Bαβ ). (C.47)
Integral term of (C.38). The integral term in (C.38) is given by
 Λ
[ [ M Λ D− pγ ] Γ D+ φ γ ]dΓ . (C.48)
0

We need
[ [ M Λ D− pγ ] Γ D+ φ γ ] = −(η− + D+ )[ [ M Λ D− pγ ] Γ φ γ ]
= −(η− + D+ )[ (D− + χ+ )[ M Λ pγ ] Γ φ γ ]
= (η− + D+ )(η+ + χ+ )[ [ M Λ pγ ] Γ φ γ ]. (C.49)
Using (C.42) we get
[ φγ Γ [ M Λ pγ ] ] = −i[ φ γ Γ [ pγ Λ Kαβ ] ]Bαβ + i(D− + η+ + χ+ )[ φ γ Γ Kγ α ] D + φ α
− i(D+ + η− + χ− )[ φ γ Γ Kγ α ]D− φ α . (C.50)
 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2275

γ
We have [ φ γ Γ Kγ α ] = Γγ α = 0, so

[ [ M Λ pγ ] Γ φ γ ] = i[ φ γ Γ [ pγ Λ Kαβ ] ]Bαβ , (C.51)

and

[ [ M Λ D− pγ ] Γ D+ φ γ ] = iη− η+ [ φ γ Γ [ pγ Λ Kαβ ] ]Bαβ , (C.52)

so the quantum correction is

λ[ φ γ Γ [ pγ Λ Kαβ ] ]Bαβ . (C.53)

Using Jacobi, this is

γ
λ[ pγ Λ [ φγ Γ Kαβ ] ]Bαβ = −iλ[ pγ Λ Γαβ ]Bαβ . (C.54)

In total. Thus, (C.38) is the sum of (C.44), (C.47) and (C.54):

γ
χ+ D+ M − χ− χ+ M + χ− (Γαβ Bαβ D− pγ − iD− Kαβ Bαβ ) − iχ+ ([ pγ Λ Kαβ ]Bαβ D+ φ γ )
γ γ
+ D+ (Γαβ Bαβ )D− pγ − iD− D+ (Kγ α D− φ α )D+ φ γ − iD− ([ pγ Λ Kαβ ]Bαβ )D+ φ γ − (λ + ∂)(i [ pγ Λ Γαβ ]Bαβ ). (C.55)

We have

i[ pγ Λ Kαβ ]Bαβ D+ φ γ = iE+ , (C.56)

γ
Γαβ Bαβ D− pγ − iD− Kαβ Bαβ = iE− . (C.57)

Also,

− iD− D+ (Kγ α D− φ α )D+ φ γ = −iD+ (D− Kαβ Bαβ ) + i2∂ Kαβ Bαβ + i2Kαβ D+ φ α ∂ D− φ β , (C.58)

so

[ M Λ G0+ ] = −χ− χ+ M + χ+ D+ M + i(χ− + D+ )E− − i(χ+ + D− )E+

γ
+ 2i∂ Kαβ Bαβ + 2iKαβ D+ φ α ∂ D− φ β − Γαβ Bαβ D+ D− pγ + i[ pγ Λ Kαβ ]Bαβ D− D+ φ γ
γ
− (λ + ∂)(i [ pγ Λ Γαβ ]Bαβ ). (C.59)

[ G0+ Λ M ], and taking the sum. Using skew-symmetry, from (C.59), we calculate [ G0+ Λ M ]:
[ G0+ Λ M ] = χ− χ+ M + 2λM + 2∂ M + χ− D− M + iχ− E− − iχ+ E+ − 2i∂ Kαβ Bαβ − 2iKαβ D+ φ α ∂ D− φ β
γ γ
+ Γαβ Bαβ D+ D− pγ − i[ pγ Λ Kαβ ]Bαβ D− D+ φ γ − λi [ pγ Λ Γαβ ]Bαβ . (C.60)

Taking the sum of (C.59) and (C.60), we have

[ G0+ Λ M ] + [ M Λ G0+ ] = (2λ + 2∂ + χ− D− + χ+ D+ )M + i(2χ− + D+ )E−

γ
− i(2χ+ + D− )E+ − (2λ + ∂)(i [ pγ Λ Γαβ ]Bαβ ). (C.61)

C.2.3. Summing the results from C.2.1 and C.2.2

We want to show that (C.8a) is fulfilled. From (C.37) and (C.61), we get

[ G0+ Λ M ] + [ M Λ G0+ ] + [ M Λ M ] = (2λ + 2∂ + χ+ D+ + χ− D− ) M

γ γ γ
  
− (2λ + ∂) −Kγ δ [ Γαβ Γ φ δ ] + iΓαδ Γγδβ + i[ pγ Λ Γαβ ] Bαβ . (C.62)

The parenthesis of the last line is

γ γ γ γ γ
− Kγ δ [ Γαβ Γ φ δ ] + iΓαδ Γγδβ + i[ pγ Λ Γαβ ] = iΓαδ Γγδβ − i∂γ Γαβ = 0. (C.63)

This is zero in the coordinates chosen. So, (C.8a) is fulfilled. The corresponding equation for the −-sector comes by
exchanging + and −. We have thus shown that

d
[ G± Λ G± ] = (2λ + 2∂ + χ+ D+ + χ− D− ) G± + λχ1 χ2 . (C.64)
2
We now want to show that G+ and G− commute.
2276 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

C.2.4. [ G0− Λ M ] − [ M Λ G0+ ]

We want to calculate [ G0− Λ M ] − [ M Λ G0+ ]. From (C.60) we see that [ G0− Λ M ] is

[ G0− Λ M ] = χ+ χ− M + 2λM + 2∂ M + χ+ D+ M + iχ+ E+ − iχ− E− − 2i∂ Kαβ Bαβ + 2iKαβ D− φ α ∂ D+ φ β

γ γ
+ Γαβ Bαβ D− D+ pγ − i[ pγ Λ Kαβ ]Bαβ D+ D− φ γ + λi [ pγ Λ Γαβ ]Bαβ . (C.65)
Note that when we exchange + and −, or equivalently 1 and 2 in the numbering of the supersymmetries, we keep the
integration order in the integrals fixed. This yields an extra minus sign in the quantum term above.
The difference between (C.65) and (C.59) is
[ G0− Λ M ] − [ M Λ G0+ ] = (χ+ χ− + χ− χ+ + 2λ + 2∂)M + i(2χ+ + D− )E+ − i(2χ− + D+ )E−
− 4i∂ Kαβ Bαβ − 2iKαβ D+ φ α ∂ D− φ β − 2iKαβ ∂ D+ φ α D− φ β
γ
+ Γαβ Bαβ (D− D+ + D+ D− )pγ − i[ pγ Λ Kαβ ]Bαβ (D− D+ + D+ D− )φ γ
γ
+ i(2λ + ∂) ([ pγ Λ Γαβ ]Bαβ )
= +i(2χ+ + D− )E+ − i(2χ− + D+ )E− + 2∂ M − 4i∂ Kαβ Bαβ − 2iKαβ ∂ Bαβ
γ γ
+ 2Γαβ Bαβ ∂ pγ − 2i[ pγ Λ Kαβ ]Bαβ ∂φ γ + i(2λ + ∂) ([ pγ Λ Γαβ ]Bαβ ). (C.66)
The third line of (C.66) can be simplified, noting that
γ γ γ
Γαβ Bαβ ∂ pγ − i[ pγ Λ Kαβ ]Bαβ ∂φ γ = iΓαβ Bαβ Kγ σ ∂φ σ + iΓαβ Bαβ Kγ σ̄ ∂φ σ̄
σ
− iΓαβ Kσ γ Bαβ ∂φ γ + iKαβγ Bαβ ∂φ γ

= iKαβ γ̄ Bαβ ∂φ γ̄ + iKαβγ Bαβ ∂φ γ = i∂ Kαβ Bαβ . (C.67)

So, finally,
γ
[ G0− Λ M ] − [ M Λ G0+ ] = i(2χ+ + D− )E+ − i(2χ− + D+ )E− + i(2λ + ∂) ([ pγ Λ Γαβ ]Bαβ ), (C.68)
and, using (C.37) and (C.63),
γ γ γ
  
[ G0− Λ M ] − [ M Λ G0+ ] − [ M Λ M ] = +(2λ + ∂) −Kγ δ [ Γαβ Γ φ δ ] + iΓαδ Γγδβ + i [ pγ Λ Γαβ ] Bαβ

= 0. (C.69)
Thus
[ G+ Λ G− ] = 0. (C.70)

Appendix D. Derivation of the Hamiltonian of the N = (2, 2) supersymmetric sigma model

The action for an N = (2, 2) supersymmetric sigma model with a Kähler target manifold is given by

S= dσ dτ dθ+
1
dθ−
1
dθ+
2
dθ−
2
K (Φ , Φ̄ ), (D.1)

where K is the Kähler potential, and Φ = {Φ α } is a chiral superfield, and Φ̄ = {Φ ᾱ } is an anti-chiral superfield. We use
indices µ, ν, . . . to denote real coordinates, and α, β, . . . to denote complex coordinates.
We have two copies of the N = (1, 1) algebra1 :

(Di± )2 = i∂± , {Di+ , Dj− } = 0, {D1± , D2± } = 0, i, j = 1, 2, (D.2)

where
∂
Di± = + iθ±i ∂± , ∂± = ∂0 ± ∂1 . (D.3)
∂θ±i
For chiral and anti-chiral superfields, the two supersymmetries are related:

D1± Φ α = iD2± Φ α , D1± Φ ᾱ = −iD2± Φ ᾱ . (D.4)

1 Here we misuse the spinor notation. For example, the partial derivative ∂ should be understood as ∂
+ ++ in spinor indices. Since we are after the
Hamiltonian treatment, the Lorentz covariance is not the issue.
 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278 2277

In the physics literature Di± are typically combined into complex operators and also it is more customary to use the complex
θ ’s (for example, see the conventions in [17]).
We now want to go to Hamiltonian formalism. We are going to integrate out two odd coordinates, and write (D.1) in a
first order form.
Let us define a new set of θ ’s:
θ0  θ 1 
 1 1 1 −i i +
θ 2  1 −i i 1 1  θ 1
.
 0 −
 1 = √   (D.5)
θ1  −1 −i  θ+2 

2 1 i

θ12 i −i 1 −1 θ−2
1
dσ dτ dθ11 dθ12 dθ01 dθ02 K . We now want to integrate out θ01 and θ02 . Introduce new differential

The action then is S = − 2
operators:
1 1
D10 = √ (D1+ + D1− ), D20 = √ (D2+ + D2− ), (D.6)
2 2
1 1
D11 = √ (D1+ + iD2− ), D21 = √ (D2+ + iD1− ). (D.7)
2 2
We have
∂ 1 1
D10 = + iθ01 ∂0 + (iθ11 − θ12 )∂0 + (iθ11 + θ12 )∂1 , (D.8)
∂θ01 2 2
∂ 1 1
D20 = + iθ02 ∂0 + (iθ12 − θ11 )∂0 + (iθ12 + θ11 )∂1 , (D.9)
∂θ02
2 2
and
(D10 )2 = (D20 )2 = i∂0 , (D11 )2 = (D21 )2 = i∂1 . (D.10)
Under integration,

1
S=− dσ dτ dθ11 dθ12 D10 D20 K |θ 1 =θ 2 =0 . (D.11)
2 0 0

Now,

D10 D20 K = Kµν D10 Φ µ D20 Φ ν + Kµ D10 D20 Φ µ . (D.12)

Due to (D.4), we have

D20 Φ α = −iD10 Φ α , D20 Φ ᾱ = +iD10 Φ ᾱ , (D.13)

and
Kµ D10 D20 Φ µ = −iKα D10 D10 Φ α + iKᾱ D10 D10 Φ ᾱ
= Kα ∂0 Φ α − Kᾱ ∂0 Φ ᾱ
= 2Kα ∂0 Φ α + total derivative. (D.14)
Also,
Kµν D10 Φ µ D20 Φ ν = Kµα D10 Φ µ D20 Φ α + Kµᾱ D10 Φ µ D20 Φ ᾱ
= −iKβ̄α D10 Φ β̄ D10 Φ α + iKβ ᾱ D10 Φ β D10 Φ ᾱ

= −2iKβ̄α D10 Φ β̄ D10 Φ α . (D.15)

Using (D.4) again, we note that
1 1
D10 Φ α = √ (D1+ + D1− )Φ α = √ (D1+ + iD2− )Φ α = D11 Φ α , (D.16)
2 2
1 1
D10 Φ ᾱ = √ (D1+ + D1− )Φ ᾱ = √ (−iD2+ + D1− )Φ ᾱ = −iD21 Φ ᾱ , (D.17)
2 2
so,

Kµν D10 Φ µ D20 Φ ν = −2Kβ̄α D21 Φ β̄ D11 Φ α , (D.18)

2278 J. Ekstrand et al. / Journal of Geometry and Physics 62 (2012) 2259–2278

and
 
 
S= d σ dθ
2 1 2
1d 1 θ Kβ̄α D21 Φ β̄ D11 Φ α − Kα ∂ 0 Φ α
. (D.19)


θ01 =θ02 =0
√ √
Denote θ 1 ≡ iθ11 , θ 2 ≡ iθ12 and ∂ ≡ ∂1 . Let
√ ∂
D1 ≡ −i iD11 |θ 1 =θ 2 =0 = + θ 1 ∂, (D.20)
0 0 ∂θ 1
√ ∂
D2 ≡ −i iD21 |θ 1 =θ 2 =0 = 2 + θ 2 ∂, (D.21)
0 0 ∂θ
and φ µ ≡ Φ µ |θ 1 =θ 2 =0 . Then
0 0
  
S = d2 σ dθ 2 dθ 1 iKα ∂0 φ α − Kβ̄α D2 φ β̄ D1 φ α
  
1
= d2 σ dθ 2 dθ 1 iKα ∂0 φ α − H , (D.22)
2
with

H = Kβ̄α D2 φ β̄ D1 φ α − Kβ̄α D1 φ β̄ D2 φ α . (D.23)

From (D.22) we see that the Hamiltonian density is given by (D.23). The momenta are
pα = iKα , pᾱ = 0. (D.24)
The definitions of the momenta give the second class constraints pα − iKα = 0 and pᾱ = 0, leading to the Dirac brackets

{φ α , φ β̄ }∗ = ωαβ̄ , {φ α , pβ }∗ = δβα , {φ ᾱ , pβ }∗ = iωᾱα Kαβ , (D.25)

with the remaining brackets being zero.
Let us define new combinations of the derivatives D1 and D2 :
1
D± ≡ √ (D1 ∓ iD2 ). (D.26)
2
We can then write the Hamiltonian (D.23) as

H = Gc− − Gc+ , (D.27)

with
Gc± = D∓ pα D± φ α + iKαβ D± φ α D∓ φ β . (D.28)

References

[1] F. Malikov, V. Schechtman, A. Vaintrob, Chiral de Rham complex, Comm. Math. Phys. 204 (2) (1999) 439–473. arXiv:math/9803041.
[2] D. Ben-Zvi, R. Heluani, M. Szczesnya, Supersymmetry of the chiral de Rham complex, Compos. Math. 144 (02) (2008) 503–521.
[3] R. Heluani, Supersymmetry of the chiral de Rham complex II: commuting sectors, Int. Math. Res. Not. IMRN (6) (2009) 953–987. arXiv:0806.1021
[math.QA].
[4] R. Heluani, M. Zabzine, Generalized Calabi–Yau manifolds and the chiral de Rham complex, Adv. Math. 223 (2010) 1815–1844. arXiv:0812.4855
[math.QA].
[5] J. Ekstrand, R. Heluani, J. Källén, M. Zabzine, Chiral de Rham complex on special holonomy manifolds, arXiv:1003.4388 [hep-th].
[6] F. Malikov, Lagrangian approach to sheaves of vertex algebras, Comm. Math. Phys. 278 (2) (2008) 487–548.
[7] J. Ekstrand, R. Heluani, J. Källén, M. Zabzine, Non-linear sigma models via the chiral de Rham complex, Adv. Theor. Math. Phys. 13 (4) (2009) 1221–1254.
arXiv:0905.4447 [hep-th].
[8] M. Kapranov, E. Vasserot, Vertex algebras and the formal loop space, Publ. Math. Inst. Hautes Études Sci. (100) (2004) 209–269.
[9] P. Deligne, J.W. Morgan, Notes on supersymmetry (following Joseph Bernstein), in: Quantum Fields and Strings: A Course for Mathematicians, Vols.
1, 2 (Princeton, NJ, 1996/1997), Amer. Math. Soc., Providence, RI, 1999, pp. 41–97.
[10] V.G. Kac, Vertex Algebras for Beginners, second ed., in: University Lecture Series, vol. 10, American Mathematical Society, 1998.
[11] R. Heluani, V.G. Kac, Supersymmetric vertex algebras, Comm. Math. Phys. 271 (1) (2007) 103–178.
[12] B.H. Lian, A.R. Linshaw, Chiral equivariant cohomology. I, Adv. Math. 209 (1) (2007) 99–161.
[13] I. Vaisman, Lectures on the Geometry of Poisson Manifolds, in: Progress in Mathematics, Birkhäuser Verlag, 1994.
[14] M. Zabzine, Hamiltonian perspective on generalized complex structure, Comm. Math. Phys. 263 (3) (2006) 711–722.
[15] A. Bredthauer, U. Lindström, J. Persson, M. Zabzine, Generalized Kähler geometry from supersymmetric sigma models, Lett. Math. Phys. 77 (2006)
291–308. arXiv:hep-th/0603130.
[16] K. Hori, S. Katz, A. Klemm, R. Pandharipande, R. Thomas, C. Vafa, R. Vakil, E. Zaslow, Mirror Symmetry, in: Clay Mathematics Monographs, vol. 1,
American Mathematical Society, Providence, RI, 2003.
[17] U. Lindström, M. Roček, R. von Unge, M. Zabzine, Generalized Kähler manifolds and off-shell supersymmetry, Comm. Math. Phys. 269 (2007) 833–849.
arXiv:hep-th/0512164.