Bogoliubov Transformation and Schrodinger Representation on Curved Space

Technical Summary: Bogoliubov Transformation and Schrodinger Representation on Curved Space

Problem Statement
In quantum field theory (QFT) on curved spacetime, it is conventionally accepted that the dynamics described by the Schrödinger equation, which evolves states between Cauchy surfaces, should be unitary. However, previous studies have demonstrated that this expectation fails when one attempts to describe dynamics on a fixed Hilbert space, particularly when the chosen Cauchy surfaces are not preserved by a timelike Killing vector field. This issue persists even in Minkowski space for arbitrary Cauchy slices. The core difficulty lies in the fact that different choices of complex structure $J$ on the phase space (which define the vacuum and the Hilbert space) are generally not unitarily equivalent. Specifically, the condition for unitary equivalence between two Hilbert spaces $H_{J_1}$ and $H_{J_2}$ requires the difference $J_1 - J_2$ to satisfy the Hilbert-Schmidt condition, a condition often violated in curved spacetime settings.

Agullo and Ashtekar (Ref. [18]) previously addressed this by proposing that Schrödinger dynamics does not occur on a fixed Hilbert space but rather on a "Hilbert bundle" where the fiber (the Hilbert space) changes with time. They postulated a "Generalized Unitarity condition" (Eq. 2 in the paper) which allows for a unitary evolution operator $U$ even when the initial and final Hilbert spaces are not kinematically equivalent, provided the evolution respects the dynamical change of the complex structure.

Methodology
This paper aims to explicitly construct the evolution operator $U$ satisfying the Generalized Unitarity condition by deriving the Schrödinger equation directly, rather than inferring it from the classical picture as done in Ref. [18]. The authors employ the following methodological framework:

1. Classical Setup: The authors consider a linear real Klein-Gordon field on a globally hyperbolic spacetime. They construct the canonical phase space $\Gamma$ and demonstrate that the symplectic structure is translation invariant.
2. Geometric Quantization: Following the formalism of geometric quantization (Refs. [22, 23]), the authors quantize the canonical phase space. They introduce a complex structure $J$ compatible with the symplectic form to define the Hilbert space $H_J$ of holomorphic wavefunctions.
3. Bogoliubov Transformation and Connection: A key technical step involves handling the time evolution of the complex structure. The authors utilize a connection term $A$ (Eq. 32) derived from the change in complex structure $\delta J$ along the Hamiltonian flow. This connection implements the Bogoliubov transformation required to map states between Hilbert spaces defined by different complex structures.
4. Hamiltonian Construction: The authors construct a Hamiltonian operator $\hat{H}$ (Eq. 54) that includes both the pre-quantum operator $Pr(H)$ and a specific connection term involving the change in complex structure. This Hamiltonian is designed to preserve the holomorphicity of wavefunctions during time evolution.
5. Schrödinger Equation: The time evolution of the state is governed by the Schrödinger equation $i\partial_t \psi = \hat{H}\psi$ (Eq. 50). The finite time evolution operator $U(t_f, t_i)$ is defined as the time-ordered exponential of this Hamiltonian.

Key Contributions and Results

• Explicit Construction of the Evolution Operator: The paper derives the explicit form of the operator $U(t_f, t_i)$ that propagates states between arbitrary Cauchy surfaces. This operator maps the Hilbert space $H_{J(t_i)}$ to $H_{J(t_f)}$, where the complex structure $J$ evolves naturally with time.
• Derivation of Generalized Unitarity: By analyzing the action of the evolved operator on the vacuum state, the authors derive the Generalized Unitarity condition (Eq. 60) proposed by Agullo and Ashtekar. They show that:
  $$\langle \Psi' | U(t_f, t_i)^{-1} \hat{\hat{\phi}}_i U(t_f, t_i) | \Psi \rangle = \langle \Psi' | E_i^j \hat{\phi}_j | \Psi \rangle$$
  where $\hat{\hat{\phi}}$ is quantized with respect to the final complex structure $J(t_f)$, $\hat{\phi}$ is quantized with respect to the initial structure $J(t_i)$, and $E$ is the classical time evolution map.
• Proof of Unitarity via Hilbert-Schmidt Condition: The authors prove that the evolution operator $U$ is unitary if and only if the operator $[J(t_f) - E J(t_i) E^{-1}]$ satisfies the Hilbert-Schmidt condition (Eq. 64).
  • Crucially, because the authors define the time evolution of the complex structure as the natural flow generated by the Hamiltonian ($J(t_f) = E J(t_i) E^{-1}$), the difference term vanishes. Consequently, the Hilbert-Schmidt condition is trivially satisfied, and the dynamics are always unitary within this framework.
• Distinction from Kinematical Unitarity: The paper clarifies that while the dynamical evolution is unitary, the kinematical identification between the initial and final Hilbert spaces (i.e., comparing $J(t_i)$ and $J(t_f)$ directly without the evolution map $E$) is generally not unitary. The connection term $A$ alone is not unitary in infinite dimensions; it is the specific combination of the Hamiltonian flow and the connection that ensures unitarity.

Significance and Claims
The paper claims to provide a constructive derivation of the Schrödinger picture for QFT on curved spacetime, complementing the work of Agullo and Ashtekar. By explicitly incorporating the Bogoliubov transformation into the Schrödinger equation via a connection term, the authors demonstrate that quantum dynamics can be formulated as a unitary process on a Hilbert bundle.

The significance lies in resolving the "unitarity puzzle" by showing that the failure of unitarity in previous attempts was due to the insistence on a fixed Hilbert space. By allowing the Hilbert space to evolve (specifically, by evolving the complex structure naturally), the authors show that a unitary Schrödinger dynamics exists for arbitrary Cauchy surfaces. The paper modestly notes that its analysis relies on translation-invariant complex structures and a finite-dimensional notation limit, leaving the rigorous functional analytic treatment of infinite-dimensional operator domains for future work.