Contact reduction

From MaRDI portal



Abstract: In this article I propose a new method for reducing a co-oriented contact manifold M equipped with an action of a Lie group G by contact transformations. With a certain regularity and integrality assumption the contact quotient Mmu at muinfg∗ is a naturally a co-oriented contact orbifold which is independent of the contact form used to represent the given contact structure. Removing the regularity and integrality assumptions and replacing them with one concerning the existence of a slice, which is satisfied for compact symmetry groups, results in a contact stratified space; i.e, a stratified space equipped with a line bundle which, when restricted to each stratum, defines a co-oriented contact structure. This extends the previous work of the author and E. Lerman [LW].


The symplectic reduction is well-understood through the works of Kazhdan-Konstant-Sternberg, Marsden-Weinstein-Meyer and Lerman-Sjamaar [see \textit{R. Sjamaar} and \textit{E. Lerman}, Ann. Math. (2) 134, 375-422 (1991; Zbl 0759.58019)]. The contact reduction for contact manifolds is a parallel theory of the symplectic reduction for symplectic manifolds. Symplectic and contact manifolds are twin, can be related to each other through symplectization and invariant hypersurface techniques. Examples in section 2 of the paper give such relations. There are several earlier works on contact reduction: (1) Using the intertwining relation between symplectic and contact structures, Guillemin-Sternberg contact reduction is defined via co-isotropic reduction; (2) by Marsden-Weinstein-Meyer quotient, Lerman and Willett defined another contact reduction. (Remark: (1) and (2) are different, but related in Proposition 5.8 of the paper under review); (3) C. Albert, H. Geiges and F. Loose independently gave a contact reduction. Albert's definition works for all values \(\mu \in \mathfrak{g}^*\) but depends upon the contact form (see Appendix of the paper). The paper under review extends the author's previous work with Lerman, and shows that contact reduction carries a contact structure on each stratum as in the symplectic reduction in the paper of Lerman and Sjamaar mentioned before. Let a Lie group \(G\) act by contact transformation on a contact manifold \((M, \alpha)\), where \(\alpha \) is the contact 1-form. The contact moment map \(\Phi_{\alpha}: M \to {\mathfrak{g}}^*\) associated to the contact action is defined by \(\langle \Phi_{\alpha}(x), A \rangle = \alpha_x (A_M(x))\), for \(x\in M\) and \(A \in {\mathfrak{g}}\). Through the relation between symplectic and contact structures, the contact moment map and symplectic moment map are naturally related by identification and restriction. The Reeb flow preserves the level sets of \(\Phi_{\alpha}\). The notable difference is that a point of \(\Phi_{\alpha}\) may be critical yet have a discrete stabilizer. Note that a regular point of the symplectic moment map has a discrete stabilizer. Define the kernel group \(K_{\mu}\) for \(\mu \in {\mathfrak{g}}^*\) to be the unique connected Lie subgroup of \(G_{\mu}\) with the Lie algebra \(\ker \mu|_{{\mathfrak{g}_{\mu}}}\). To guarantee the existence of \(K_{\mu}\) certain commensurability is required for \(G\) (this is not stated in the paper). The contact reduction of \((M, \alpha)\) by \(G\) at \(\mu\) is defined by \[ M_{\mu} = \Phi_{\alpha}^{-1}(R^+\mu)/K_{\mu}. \] (i) \(M_{\mu}\) is topologically independent of the contact form \(\alpha\). (ii) \(\Phi_{\alpha}\) is transversal to \(R^+\mu\) if and only if the kernel group \(K_{\mu}\) acts locally free on the level set \(\Phi_{\alpha}^{-1}(R^+\mu)\). One can replace the property `transversal' by the existence of a convex, \(R^+\)-invariant slice for \(\mu \in {\mathfrak{g}}^*\) (as it is true for compact Lie groups \(G\)). Then theorem 3.11 (result of Lerman and Willett) shows that \(M_{\mu}\) is a topologically stratified space. (iii) The kernel group \(K_{\mu}\) may not act properly on \(\Phi_{\alpha}^{-1}(R^+\mu)\) and \(K_{\mu}\) may be the isotropic subgroup of \(\mu\). These two problems damage the Hausdorff property and the contact structure on the quotient (or the contact reduction). An explicit example of this phenomenon is given in Example 3.7. Hence the hypothesis for the main results of the paper targets to avoid this situation. Hence Theorems 2 and 3 of the paper show that, under the assumptions of the existence of a convex \(R^+\)-invariant slice and of the properly action of the kernel group \(K_{\mu}\) on \(\Phi_{\alpha}^{-1}(R^+\mu)\), the contact reduction \(M_{\mu}\) is a stratified space, and each stratum carries an induced contact structure through a freely acting symmetry group. This gives an analogue of the symplectic reduction of the paper of Lerman and Sjamaar.











This page was built for publication: Contact reduction

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3151229)