Inspired by the Clebsch optimal control problem, we introduce a new variational principle that is suitable for capturing the geometry of relativistic field theories with constraints related to a gauge symmetry. Its special feature is that the Lagrange multipliers couple to the configuration variables via the symmetry group action. The resulting constraints are formulated as a condition on the momentum map of the gauge group action on the phase space of the system. We discuss the Hamiltonian picture and the reduction in the gauge symmetry by stages in this geometric setting. We show that the Yang–Mills–Higgs action and the Einstein–Hilbert action fit into this new framework after a (1 + 3)-splitting. Moreover, we recover the Gauß constraint of Yang–Mills–Higgs theory and the diffeomorphism constraint of general relativity as momentum map constraints.
REFERENCES
For example, we may take T*Q to be TQ and the pairing given by a Riemannian structure on Q. In applications, the fiber of TQ is often a space of mappings so that a convenient choice of the cotangent bundle consists of regular distributions inside the space of all distributions.
More generally, one could give up the assumption of the existence of a local addition and instead work in the diffeological category.23
That is, .
Here, we have used the identity , which in infinite dimensions requires additional assumptions of functional-analytic nature. For a precise formulation of the bifurcation lemma in an infinite-dimensional context, see Ref. 12.
To be more precise, the reduction by stages theorem Ref. 26 (Theorem 4.2.2) is formulated for the free and proper action of a finite-dimensional group on a finite-dimensional phase space. Nonetheless, the peculiarities of our infinite-dimensional setting can be handled easily due to the simple form of the first reduction.
This one-to-one correspondence will be used throughout the text without further notice.
Although this convention is a bit nonstandard, the consistent use of dual-valued forms has the advantage that dual objects are clearly marked, which will be helpful later to identify them as points in the cotangent bundle (as opposed to elements of the tangent bundle).
The diamond operator often occurs in the study of Lie–Poisson systems, and this is where we borrowed the notation from. Note that in a purely algebraic setting, the diamond product boils down to a map dual to the Lie algebra action. Hence, it is a momentum map for the G-action on T*F. For example, if we consider the action of G = SO(3) on and identify F* with , then the diamond product becomes the classical cross product.