Remarks on A-branes, mirror symmetry, and the Fukaya category
Introduction
Let X be a weak Calabi–Yau manifold, i.e. a complex manifold with c1(X)=0 which admits a Kähler metric. Given a Ricci-flat Kähler metric G on X, and a B-field (a class in ), one can canonically construct an N=2 supersymmetric sigma-model with “target” X. On physical grounds, the quantized version of this model has N=2 superconformal symmetry and describes propagation of closed strings on X. In this note we set B=0 for simplicity. According to Calabi’s conjecture proved by Yau, we can parametrize G by the cohomology class of its Kähler form ω. A weak Calabi–Yau manifold equipped with a Kähler form ω will be called a physicist’s Calabi–Yau.
It sometimes happens that two different physicist’s Calabi–Yau manifolds (X,ω) and (X′,ω′) give rise to a pair of N=2 superconformal field theories (SCFTs) related by a mirror morphism [1], [2]. A mirror morphism of N=2 SCFT is an isomorphism of the underlying N=1 SCFT which acts on the N=2 super-Virasoro algebra as a mirror involution [3], [4]. In this case, one says that (X,ω) and (X′,ω′) are mirror to each other. (For a concise explanation of the notions involved and further references, see [5]. An algebraically minded reader may find it useful to consult Ref. [6] for a careful definition of N=2 SCFTs and their morphisms.)
A long-standing problem is to understand the mirror relation from a mathematical viewpoint, i.e. without a recourse to the ill-defined procedure of quantizing a sigma-model. A fascinating conjecture has been put forward by Kontsevich [7]. He observed that to any physicist’s Calabi–Yau (X,ω), one can associate two triangulated categories: the well-known bounded derived category of coherent sheaves Db(X) and the still mysterious Fukaya category . Objects of the category Db(X) are bounded complexes of coherent sheaves. Objects of the Fukaya category are (roughly speaking) vector bundles on Lagrangian submanifolds of X equipped with unitary flat connections. The Homological Mirror Symmetry Conjecture (HMSC) asserts [7] that if two algebraic physicist’s Calabi–Yau manifolds (X,ω) and (X′,ω′) are mirror to each other, then Db(X) is equivalent to , and is equivalent to Db(X′). So far this conjecture has been proved only for elliptic curves [8].
From a physical viewpoint, complexes of coherent sheaves are D-branes of the topological B-model (B-branes). We remind that the B-model of a physicist’s Calabi–Yau (X,ω) is a topological “twist” of the corresponding N=2 SCFT [9]. The twisted theory is a two-dimensional topological field theory whose correlators do not depend on ω. Morphisms between the objects of Db(X) are identified with the states of the topological string stretched between pairs of B-branes, and the compositions of morphisms are computed by the correlators of the B-model. This correspondence has been intensively discussed in the physics literature (see e.g. Refs. [10], [11], [12], [13] and references therein), and will be taken as a starting point here.
An N=2 SCFT has another twist, called the A-twist [9]. The corresponding topological field theory (the A-model) is insensitive to the complex structure of X, but depends non-trivially on the symplectic form ω. D-branes of the A-model are called A-branes. Mirror morphisms exchange A- and B-twists and A- and B-branes. Thus from a physical viewpoint the mirror of Db(X) is the category of A-branes on X′.
It can be shown that any object of the Fukaya category gives rise to an A-brane. Moreover, the recipe for computing morphisms between such A-branes can be derived heuristically in the path integral formalism, and it reproduces the definition of morphisms in the Fukaya category [14]. Therefore, the majority of researchers in the field assumed that the mirror relation between the categories of A- and B-branes is essentially a restatement of the HMSC in physical terms.1
In this note, we will argue that this is not the case, because A-branes are not necessarily Lagrangian submanifolds in X. This was mentioned already in one of the first papers on the subject [15], but the general conditions for a D-brane to be an A-brane have not been determined there. In Section 3, we will show that a coisotropic submanifold of X with a unitary line bundle on it is an A-brane if the curvature of the connection satisfies a certain algebraic condition. We remind that a submanifold Y of a symplectic manifold (X,ω) is called coisotropic if the skew-complement of TY⊂TX|Y with respect to ω is contained in TY. In the physical language, a coisotropic submanifold is a submanifold locally defined by first-class constraints. One can easily see that the dimension of a coisotropic submanifold is at least half the dimension of X, and that a middle-dimensional coisotropic submanifold is the same thing as a Lagrangian submanifold. Thus we show that the category of A-branes contains, besides Lagrangian A-branes, A-branes of larger dimension.
In Section 4, we explore the geometric interpretation of the algebraic condition on the curvature of the line bundle. We will see that an A-brane is naturally a foliated manifold with a transverse holomorphic structure. The notion of transverse holomorphic structure is a generalization of the notion of complex structure to foliated manifolds. If the space of leaves of a foliated manifold Y is a smooth manifold, a transverse holomorphic structure on Y is simply a complex structure on the space of leaves. The general definition is given in Section 4. In addition to being transversely holomorphic, a coisotropic A-brane also carries a transverse holomorphic symplectic form.
In the case of a Lagrangian A-brane, the foliation has codimension zero, there are no transverse directions, and the transverse holomorphic structure is not visible. In general, the foliation is determined by the restriction of ω to Y, while the transverse holomorphic structure comes from the curvature of the line bundle on the brane.
Interestingly, to prove that an A-brane has a natural transverse holomorphic structure, one needs to use some facts from bihamiltonian geometry. The subject matter of bihamiltonian geometry is manifolds equipped with two compatible (in a sense explained below) Poisson structures. In our case, the underlying manifold is foliated, and one is dealing with transverse Poisson structures. (If the space of leaves is a manifold, specifying a transverse Poisson structure is the same as specifying an ordinary Poisson structure on the space of leaves.) One of the transverse Poisson structures arises from the symplectic form ω in the ambient space X, and the other one from the curvature of the line bundle on Y.
Our understanding of the category of A-branes is far from complete. Nevertheless, it is clear that generally it includes objects other than Lagrangian submanifolds with flat vector bundles. (There are certain special, but important, cases where there seem to be no non-Lagrangian A-branes, like the case of an elliptic curve, or a simply connected Calabi–Yau 3-fold.) Therefore, the Fukaya category must be enlarged with coisotropic A-branes for the HMSC to be true. (This is somewhat reminiscent of the remark made in Ref. [7] that Lagrangian foliations may need to be included in the Fukaya category.) This is discussed in more detail in Section 5.
Since our arguments are ultimately based on non-rigorous physical reasoning, a skeptic might not be convinced that the HMSC needs serious modification. To dispel such doubts, we discuss in Section 2 mirror symmetry for tori and show that under mild assumptions the usual Fukaya category cannot capture the subtle behavior of Db(X) under the variation of complex structure. Inclusion of coisotropic A-branes seems to resolve the problem.
Section snippets
Why Lagrangian submanifolds are not enough
In this section, we give some examples which show that the Fukaya category must be enlarged with non-Lagrangian objects for the HMSC to be true. We will exhibit a mirror pair of tori such that mirror symmetry takes a holomorphic line bundle (a B-brane) on the first torus to a complex line bundle on the second torus. This means that the latter line bundle is an A-brane.
It is well known that the derived category of coherent sheaves behaves in a very non-trivial manner under a variation of complex
World-sheet approach to A-branes
This section assumes some familiarity with supersymmetric sigma-models (on the classical level) and superconformal symmetries. Let X be a Käler manifold with metric G and Kähler form ω. The complex structure on X is given by I=G−1ω. The supersymmetric sigma-model with target X classically has (2,2) superconformal symmetry. Quantum anomaly destroys this symmetry unless c1(X)=0.
Let j:Y→X be a submanifold in X, and E be a line bundle on Y with a unitary connection. Our goal is to derive the
The geometry of A-branes
In this section, we discuss the geometry of a general coisotropic A-brane. We will see that it has some beautiful connections with bihamiltonian geometry and foliation theory.
A coisotropic submanifold Y of a symplectic manifold X has several equivalent definitions. The usual definition is that at any point p∈Y the skew-orthogonal complement of TYp is contained in TYp. Another popular definition is that Y is locally defined by first-class constraints. In other words, locally Y can be represented
A-branes and homological mirror symmetry
We have shown that an A-brane is a coisotropic submanifold in X, and that it is naturally a foliated manifold with a transverse holomorphic structure. Now let us see how this fits in with the HMSC.
As explained in Section 1, the mirror of the derived category is the category of A-branes. We have seen that in general the set of A-branes includes non-Lagrangian coisotropic branes, and therefore the Fukaya category must be enlarged with such A-branes for the HMSC to be true. Of course, in some
Acknowledgements
We are grateful to Dan Freed for useful suggestions and to Ezra Getzler for pointing out the relevance of bihamiltonian geometry. Some preliminary results have been presented by AK at the Duality Workshop, ITP, Santa Barbara, 18 June–13 July, 2001. AK would like to thank Ron Donagi, Dan Freed, Ezra Getzler, Tony Pantev, and other participants for stimulating discussions, and the organizers for making this workshop possible. AK was supported in part by DOE grants DE-FG02-90ER40542 and
References (27)
- et al.
Duality in Calabi–Yau moduli space
Nucl. Phys. B
(1990) - et al.
A pair of Calabi–Yau manifolds as an exactly soluble superconformal theory
Nucl. Phys. B
(1991) - et al.
Chiral rings in N=2 superconformal theories
Nucl. Phys. B
(1989) - L. Dixon, Lectures at 1987 ICTP Summer Workshop in High Energy Physics and...
- C. Voisin, Mirror Symmetry, American Mathematical Society, Providence, RI,...
- A. Kapustin, D. Orlov, Vertex algebras, mirror symmetry, and D-branes: the case of complex tori....
- M. Kontsevich, Homological algebra of mirror symmetry....
- A. Polishchuk, E. Zaslow, Categorical mirror symmetry: the elliptic curve, Adv. Theor. Math. Phys. 2 (1998) 443...
- E. Witten, Mirror manifolds and topological field theory, Essays on Mirror Manifolds, International Press, Hong Kong,...
- M.R. Douglas, D-branes, categories and N=1 supersymmetry....
Cited by (125)
Interfaces and quantum algebras, I: Stable envelopes
2023, Journal of Geometry and PhysicsGauge Theory and the Analytic Form of the Geometric Langlands Program
2024, Annales Henri PoincareInstances of higher geometry in field theory
2023, European Physical Journal: Special TopicsGauge Theory, Sigma Models and Generalised Geometry
2023, Proceedings of Science