We extend the quantizationà la Faddeev-Jackiw for non-autonomous singular systems. This leads to a generalization of the Schrödinger equation for those systems. The method is exemplified by the quantization of the damped harmonic oscillator and the relativistic particle in an external electromagnetic field.
PACS numbers:The quantization of constrained systems is almost as old as the beginning of quantum mechanics. It was Dirac [1] who elaborated a Hamiltonian approach with a categorization of constraints and the introduction of the socalled Dirac brackets. Later, Faddeev and Jackiw [2] suggested an alternative and generally simpler method based on a symplectic structure. Recently, we have proposed a third approach for classically soluble constrained systems where the brackets between the constants of integration are computed. This method does neither require Dirac formalism nor the symplectic method of FaddeevJackiw [3]. All three approaches were developed for autonomous constrained systems only. The quantization of non-autonomous singular systems has turned to be non trivial [4]. Gitman and Tyutin [5], via notably the introduction of a conjugate momentum of time, could extend Dirac approach and brackets for those systems. In the present work, our aim is to generalize the Faddeev-Jackiw symplectic approach to non-autonomous constrained systems. This leads to a generalization of the Schrödinger equation which encompasses non-autonomous singular systems. The quantization of a relativistic particle in an electromagnetic field is solved by this method, constituting an original derivation of the Dirac equation.Consider a non autonomous Lagrangian one-form as inwhere ξ j are the N -component phase-space coordinates with j = 1, ..., N , where a j and H are time-dependent. The Euler-Lagrange equations lead tȯfor an invertible antisymmetric matrix f ij . The dot denotes differentiation with respect to t. Eq. (2) can not be derived from Hamilton equations through brackets when ∂a j /∂t = 0 (see Eq. (17)). Canonical quantization seems compromised in this case. To solve this problem via the Faddeev-Jackiw approach we first introduce a time parameter τ such that time is promoted to a dynamically variable t = t(τ ) and ξ i = ξ i (τ ). This leads to a Lagrangian L τ , given by L τ dτ = Ldt, so that the action remains the same. L τ has a gauge invariance due to the arbitrariness of the parameter τ . We thus define a new LagrangianL τ = L τ + ω (t ′ − 1) that implements the gauge constraint t ′ = 1 (prime denotes differentiation with respect to τ ) via ω(τ ) a Lagrange multiplier seen as a new variable. Dropping a total time derivative term, one ends up with the following Lagrangian one-formwhere H(ξ, t, ω) = ω defines a new Hamiltonian. The Euler-Lagrange equations with L lead to Eq (2) and the equations ω ′ = 0 and t ′ = 1. Unlike the initial Lagrangian (1), L is autonomous with respect to τ and Eq.(4) is precisely of the form studied by Faddeev-Jackiw [2]with now ζ i the N +2-component phase space coordinates defined by ζ i =...