Geometric and algebraic reduction for singular momentum maps (Q2640900)

As the authors point out this paper touches on interesting questions in mathematical physics, symplectic and algebraic geometry. Including many basic definitions and examples they have made the paper accessible to interested readers in all these fields. After reviewing symplectic geometry and Poisson algebras they turn to the main subject of interest in the paper. These are the closed subsets \(\{V\}\) of the symplectic manifold \((P,\omega\)). The constraints are defined as smooth functions that vanish on V. The set of all constraints is an ideal I(V) in the Lie algebra \(C^{\infty}(P)\). Further, the constraints are divided into two classes: In the first class \(F(V)\) enter those functions which satisfy \(\{f,I(V)\}\subseteq I(V).\) Constraints which are not in the first class are named second class. If \(I(V)=F(V)\) then V is said to be of first class. The authors' primary interest is in constraints arising in Hamiltonian mechanics after fixing the value of the momentum map \(\mu\). When this value is regular the level set can be factorized additionally to a new symplectic manifold \((\hat V,{\hat \omega})\) [cf. \textit{J. Marsden} and \textit{A. Weinstein}, Rep. Math. Phys. 5, 121-130 (1974; Zbl 0327.58005)]. In the singular case the reduction procedure is non-unique as one faces various definitions. The authors make comparison between Dirac, geometric and algebraic reduction procedures. The main results give sufficient conditions for the geometric and algebraic reductions to agree. In particular, if the ideal \(I(\mu\)) of \(C^{\infty}\) functions on P generated by the components of \(\mu\) coincides with I(V), then the geometric and algebraic reductions coincide, too. Besides, numerous examples spread throughout the paper clarify the pathology either in the Lie group action generating the momentum map or in the geometry of its level set. Central to all these results is the construction of normal coordinates for the momentum map. The paper ends with discussion of several conjectures and open problems.