Boothby-Wang construction

From testwiki
Jump to navigation Jump to search

Template:Technical Template:Short description In contact geometry, a field in mathematics, the Boothby-Wang construction (or Boothby-Wang theorem) characterises a special class of contact manifolds (P,α) as principal S1-bundles over symplectic manifolds (M,ω) where ω defines an integral cohomology class. The proof employs the theory of connection and curvature forms.

The name Boothby-Wang construction is used sometimes only for the construction of the contact manifold starting from the symplectic form (as part of contactisation procedures), sometimes for the entire statement (i.e. also the converse construction of the symplectic manifold starting from the contact form). The resulting fibration (P,α)→(M,ω) is sometimes also called Boothby-Wang Fibration.[1][2]

This result was proved in 1959 by Boothby and Wang[3] and has been subject to several generalisations.[4][5][6]

Preliminaries

Regular contact manifolds

Consider any contact manifold (P,α); since its Reeb vector field Rα∈𝔛(P) is nowhere zero, the orbits of the flow of Rα (also called the Reeb orbits) define a regular foliation ℱRα of rank 1. The contact form α∈Ω1(P) is called regular if such foliation ℱRα is simple, i.e. if the leaf space P/ℱRα is a smooth manifold such that the projection P→P/ℱRα is a surjective submersion.

Assuming that P is connected and Rα is complete, this regularity condition implies that P→P/ℱRα is either a principal ℝ-bundle or a principal S1-bundle (since the Reeb orbits are either not periodic, or periodic with the same minimal period).

In the case when P is furthermore compact (which automatically implies that Rα is complete), since the fibres of P→P/ℱRα are closed sets of P, they must be compact as well, hence they must be circles; this means that the Reeb orbits are periodic with the same minimal period and P→P/ℱRα is a principal S1-bundle.

For these reasons, in the complete setting, other authors define a regular (or Boothby-Wang) contact form α∈Ω1(P) with a stronger condition: all Reeb orbits are required to be circles with the same minimal periods. This last condition is equivalent to saying that the Reeb vector field Rα∈𝔛(P) generates a free S1-action.[1][2]

Integral symplectic manifolds

A closed differential 2-form ω∈Ω2(M) is called integral if its induced de Rham cohomology class [ω]∈HdR2(M) lies in the image of H2(M,ℤ)→H2(M,ℝ)≅HdR2(M). Equivalently, its period group

Per(ω):={∫γω | γ:S2→M homology 2-cycle }⊂ℝ

is of the kind kℤ for some integer k∈ℤ. Notice that, depending on the normalisation conventions, one might have to add a multiplicative factor 12π.

Principal bundles and connections

A crucial step in the proof relies on the following fact: given any principal G-bundle π:P→M, there is a 1-1 correspondence between k-forms on M and k-forms on P which are basic, i.e. G-invariant and π-horizontal (vanishing on any π-vertical vector field). In particular, given β∈Ωk(P), there is a unique ω∈Ωk(M) such that β=π∗ω.

In general, a connection form on π:P→M is a form α∈Ω1(P,𝔤) with values in the Lie algebra 𝔤:=Lie(G) satisfying the G-equivariance

(Rg)∗α=g−1⋅α∀g∈G

and the condition

α(x†)=x∀x∈𝔤

where x†∈𝔛(P) denotes the fundamental vector field associated to x∈𝔤, i.e. the image by the infinitesimal action 𝔤→𝔛(P).

In the case of the abelian Lie group G=S1, we have the abelian Lie algebra 𝔤=(ℝ,+), so that the adjoint representation g−1⋅ is trivial and the G-equivariance boils down to the G-invariance (Rg)∗α=α for every g∈G; in this case, this is equivalent to ℒx†(α)=0 for every x∈ℝ.

From regular contact manifolds to integral symplectic manifolds

Let (P,α) be a compact regular contact manifold. Then π:P→M:=P/S1 is a principal S1-bundle and there is a unique integral symplectic form ω∈Ω2(M) such that dα=π∗ω.[1][2] If one adopts the stronger condition of regularity, i.e. requires that the Reeb vector field generates a free S1-action, the compactness hypothesis is not needed.[6]

This follows from the more complete statement below (using the extra fact that, since S1 is compact, any S1-action is automatically proper; then M:=P/S1 is a smooth manifold and π is a principal S1-bundle).

Let π:P→M be a principal S1-bundle over a 2n-dimensional manifold M, and let α∈Ω1(P,ℝ)=Ω1(P) be a connection form (where the coefficient space ℝ is identified with the Lie algebra of the abelian Lie group S1). Then

  • there is a unique closed differential 2-form ω∈Ω2(M) such that dα=π∗ω;
  • the class [ω]∈HdR2(M) is the image of the first Chern class c1(P)∈H2(M,ℤ) via H2(M,ℤ)→H2(M,ℝ)≅HdR2(M);
  • ω is nondegenerate (i.e. it is a symplectic form) if and only if α is a contact form.

Moreover, if this happens,

  • the fundamental vector field associated to 1∈ℝ≅Lie(S1) is the Reeb vector field of α;
  • the Reeb orbits are circles and the contact form is regular.

Proof

One checks first that dα∈Ω2(P) is basic. Indeed, it is S1-equivariant since α is so; moreover, it is horizontal since π-vertical vector field are fundamental vector field of the kind x†∈𝔛(P), for x∈ℝ=Lie(S1), hence, by the properties of a connection form,

dα(x†,⋅)=ιx†(dα)=ℒx†α−d(ιx†α)=−d(α(x†))=d(x)=0.

Then, by the lemma mentioned above, there is a unique ω∈Ω2(M) such that dα=π∗ω; since exterior derivative and pullbacks commute, ω is automatically closed.

Using Chern-Weil theory, the image via H2(M,ℤ)→H2(M,ℝ)≅HdR2(M) of the first Chern class of P can be computed precisely by choosing a connection α∈Ω1(P) and considering the induced ω∈Ω2(M), so that c1(P)=[ω].

Furthermore, α∧(dα)n=α∧π∗(ωn)≠0 (i.e. α is a contact form) if and only if ωn≠0 (i.e. ω is nondegenerate).

Last, consider the fundamental vector field R:=a(1)=1†∈𝔛(P), where a:Lie(S1)≅ℝ→𝔛(P) is the infinitesimal action. To prove that it coincides with the Reeb vector field Rα∈𝔛(P), one checks that it satisfies α(R)=1 and ιR(dα)=0. But the first condition follows from α(x†)=x for x=1, while the second condition follows from the fact that fundamental vector fields are π-vertical, hence

ιR(dα)=(π∗ω)(R,⋅)=ω(dπ(R),dπ(⋅))=0.

The Reeb orbits coincide therefore with the fibres of π:P→M, i.e. they are circles, and the contact form is regular.

From integral symplectic manifolds to regular contact manifolds

Let (M,ω) be an integral symplectic manifold. Then there is a principal S1-bundle π:P→M and a regular contact form α∈Ω1(P) such that dα=π∗ω.[1][2]

This follows from the more complete statement below.

Let M be a 2n-dimensional manifold and ω∈Ω2(M) a closed differential 2-form which is integral. Then

Moreover, if this happens,

  • the fundamental vector field associated to 1∈ℝ≅Lie(S1) is the Reeb vector field of α;
  • the Reeb orbits are circles and the contact form is regular.

Proof

The existence of a principal S1-bundle π:P→M with first Chern class [ω]∈H2(M) follows from the general classification of principal S1-bundles via H2(M,ℤ).

Picking any connection form α0∈Ω1(P,ℝ)=Ω1(P), the differential dα0∈Ω2(P) is basic by the same arguments as in the previous proof. Then, by the lemma mentioned above there is an induced closed 2-form ω0∈Ω2(M) such that dα0=π∗ω0. By definition of Chern class, one has [ω]=[ω0], i.e. ω−ω0=dβ for some β∈Ω1(M). Then α:=α0+π∗β∈Ω1(P) is by construction a connection form satisfying dα=π∗ω.

Last, again with the same arguments as in the previous proof, α is a contact form if and only if ω is nondegenerate, and the further two points hold.

Relation to geometric quantisation

A prequantum (or prequantisation) line bundle over a symplectic manifold (M,ω) is a complex line bundle L→M together with a fibrewise Hermitian metric h and a Hermitian connection ∇ such that its curvature F∇ satisfies F∇=±iω (the sign depends on the conventions). The existence of this object constitutes the first step (before choosing a polarisation) to perform geometric quantisation, and it is obstructed by the integrability condition of (M,ω).

The principal bundle (P,α)→(M,ω) from the Boothby-Wang construction is equivalent to the data of a prequantisation line bundle. Indeed, given P→M one defines L→M as the vector bundle L:=(P×ℂ)/S1 associated to the natural representation ℂ∈Rep(S1). Conversely, given (L,h)→M, one defines P→M as the principal bundle of unitary frames Fr(L,h)⊂Fr(L), with structure group the unitary group U(1)≅S1. The rest of the equivalence follows from the standard equivalence between vector bundle connections and principal connections, and the fact that the curvature of a connection form α∈Ω1(P,ℝ)=Ω1(P) is given by the Maurer-Cartan expression dα+12[α,α]=dα∈Ω2(P) (since the Lie algebra Lie(S1)=ℝ is abelian).

The Boothby-Wang construction yield therefore an equivalence between regular contact manifolds and prequantisation bundles (over a given integral symplectic manifold).

Example

Consider the free and proper action of the Lie group S1⊂ℂ on the sphere S2n+1⊂ℂn+1 by multiplication on each complex coordinate; its quotient can be identified with the complex projective space ℂℙn. Then the resulting principal S1-bundle (S2n+1,αstd)→(ℂℙn,ωFS) is a Boothby-Wang fibration, where S2n+1 is endowed with the standard contact form αstd, which is automatically regular, and ℂℙn is endowed with the Fubini-Study symplectic form ωFS, which is automatically integral.

In the case n=1 one recovers the classical Hopf fibration S3→ℂℙ1≅S2.

References

Template:Reflist