Abstract
Suppose we are given an action G × M → M. In the Cartan model, an element
can be written as
where ω ∈ Ω2(M) is a two-form invariant under G and
can be considered as a G equivariant map,
from the Lie algebra, g to the space of smooth functions on M. For each ξ ∈ g, ø (ξ) is a smooth function on M, and this function depends linearly on ξ Therefore, for each m ∈ M, the value ø (ξ(m)) depends linearly on, so we can think of ø as defining a map from M to the dual space g* of the Lie algebra of g:
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Bibliographical Notes for Chapter 9
An action of G on a symplectic manifold, M,is called Hamiltonian if there exists an equivariant moment map,: M g*,having the properties described in Section 9.1. A necessary condition for an action of G on M to be Hamiltonian is that the symplectic form, w,be G invariant; however this is far from sufficient. A number of sufficient conditions for a G action to be Hamiltonian are described in [GS], Section 26. For instance if G is compact (as we have been assuming in this monograph) a G action on M is Hamiltonian if either M is compact, or H 2 (M,R) = 0 or G is semi-simple.
Berline and Vergne are, as far as we know, the first persons to make the observation that a G-action on M is Hamiltonian if and only if w is the “form part” of an equivariant symplectic form. This observation plays an essential role in their beautiful proof of the Duistermaat- Heckmann theorem in [BV]. (We will describe this proof in Section 10.9.)
The classification of homogeneous symplectic manifolds in terms of coadjoint orbits is due to Kostant [Kol]; however, the quantum version of this result was, in some sense, first observed by Kirillov. Namely, Kirillov proved that if G is a connected unipotent Lie group there is a one-one correspondence between irreducible unitary representations of G and coadjoint orbits. This result was subsequently extended by Kostant [Ko2], Kostant-Auslander [AK], Sternberg [Stl], Duflo [Du] et al. (e.g., [Zi], [Li]) to other classes of Lie groups as well. The arsenal of techniques which are used for associating unitary representations to coadjoint orbits are known collectively as “geometric quantization theory” (see [Wo]). To a large extent these techniques are due to Kirillov [Ki], Kostant [Ko2] and Souriau [So].
One important example of minimal coupling is the following: Let Y be a manifold and 7r: P —and Y a principal G-bundle. Let X = T*Y, and let M be the fiber product of X and P (as fiber bundles over Y). M is a principal G bundle with base, X; and given a connection on P,one can pull it back to M to get a connection on M For this connection the minimal coupling form (9.11)—(9.12) is symplectic for all e (See [St2].) Moreover, Weinstein [W] observed that there is a way of defining this minimal coupling form intrinsically without recourse to connections: the product, T*P x F,is a Hamiltonian G-manifold with respect to the diagonal action of G,and its symplectic reduction is symplectomorphic to W with its minimal coupling form.
This example of minimal coupling is used in elementary particle physics to describe the “classical” motion of a subatomic particle in the presence of a Yang-Mills field: Suppose that, when the field is absent, this motion is described by a Hamiltonian, H: T*Y -and R. In the presence of a Yang-Mills field (i.e., of a connection on the bundle, P -and Y) the motion is described by the Hamiltonian, p* H, p being the fibering of M over T*Y. (See [St2] and [SU].)
Let O be a coadjoint orbit of the group, G,po a fixed base point in O and Gpo the stabilizer group of po. If G po is compact, there exists a neighborhood, U of po in g* such that, for every p E U,the coadjoint orbit through p can be reconstructed by a minimal coupling construction in which the base symplectic manifold is O and the fiber symplectic manifold is a coadjoint orbit of G po For some implications of this fact for the representation theory of compact Lie groups see [GLS].
For an insightful discussion of minimal coupling from the topological perspective we recommend the paper [GLSW] of Gotay, Lashof, Sniatycki and Weinstein. (They consider the problem which we discuss in Section 9.5, namely the problem of equipping a “twisted product” of two symplectic manifolds, F and X,with a symplectic structure, from a more general point of view than ours: They don’t assume that F is a Hamiltonian G space and that W is of the form (M x F)/G)
The Duistermaat-Heckmann theorem described here is one of several versions of Duistermaat-Heckmann (another one of which we will discuss in §10.9). A version of Duistermaat-Heckmann which is easily deducible from Theorem 9.6.3 is the following: Theorem 9.11.1 Let (M, w) be a compact 2d-dimensional Hamilto-nian G-manifold with moment map, 0: M —and g*. Then the measure on g* defined by the formula: (f: g* - * R being an arbitrary continuous function) is piecewise polynomial The measure, µDH,is called the Duistermaat-Heckmann measure. Since it is compactly supported, its Fourier transform: is a C°° function; and the version of the Duistermaat-Heckmann theorem which we will describe in Chapter 10 is a formula for computing (9.28) at “generic points”, x E g
The notion of a q-Hamiltonian G-manifold is an outgrowth of recent attempts to extend various theorems in equivariant symplectic geometry to the action of loop groups on infinite dimensional manifolds. A basic theorem of Alexeev-Malkin-Meinrenken asserts that there is an equivalence of categories between the category of (infinite dimensional) symplectic manifolds equipped with a Hamiltonian loop group action with proper moment maps, and the category of finite dimensional q-Hamiltonian G-manifolds.
A beautiful observation of Alexeev-Meinrenken is that there exists an intrinsic volume form on q-Hamiltonian G-manifolds. If dim M = 2d, one might regard wd as a candidate for such a volume form. However it is in general not non-vanishing. It can be converted into a non-vanishing form by dividing by q*Xp where X p is the character of the representation of G whose dominant weight is one-half the sum of the positive roots.
An analogue of the equivariant three form in all dimensions has recently been constructed by Alexeev,Meinrenken and Woodward based on an earlier construction of Jeffrey ([Je]).
In their study of q-Hamiltonian G-spaces, Alexeev and Meinrenken have been led to consider an entirely new kind of equivariant cohomology in which (St(M) ® S(g*))G is replaced by (St(M) ® U(g)) G where U(g) is the universal enveloping algebra of g (Recall that U(g) is a filtered algebra and the Poincaré-Birkhhoff-Witt theorem asserts that its associated graded algebra is S(g) which is = S(g*) in the presence of an invariant scalar product.)
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Guillemin, V.W., Sternberg, S., Brüning, J. (1999). Equivariant Symplectic Forms. In: Supersymmetry and Equivariant de Rham Theory. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-03992-2_9
Download citation
DOI: https://doi.org/10.1007/978-3-662-03992-2_9
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-08433-1
Online ISBN: 978-3-662-03992-2
eBook Packages: Springer Book Archive