Quantum Mechanics on Lie Groups: I. Noncommutative Fourier Transforms

Technical Summary: Quantum Mechanics on Lie Groups I: Noncommutative Fourier Transforms

Problem Statement
The paper addresses the foundational challenge of establishing a general framework for quantum mechanics on Lie groups $G$, where the configuration space is a Lie group (potentially infinite-dimensional, though the current work assumes finite dimensions). While classical dynamics on such spaces are well-described by Lie-Poisson or Euler-Arnold equations, their quantization remains subtle. A primary obstacle is the transition from "position space" (the group manifold $G$) to "momentum space." Unlike the standard case on $\mathbb{R}^n$, the momentum space dual to a Lie group is noncommutative due to the group structure.

Two competing definitions of momentum space exist:

1. Discrete/Representation-theoretic: Based on the Peter-Weyl theorem, where momentum space is the set of irreducible representations (discrete for compact groups). This is natural for Fourier series but ill-suited for Wigner functions and path integrals, which require continuous phase space coordinates.
2. Continuous/Geometric: The dual of the Lie algebra, $\mathfrak{g}^*$. This provides a continuous phase space $T^*G \cong G \times \mathfrak{g}^*$, essential for semiclassical limits and path integrals, but complicates the definition of Fourier transforms due to noncommuting momenta.

The paper focuses on the second approach, aiming to construct a unitary, invertible Fourier transform mapping square-integrable wave functions on $G$ to wave functions on $\mathfrak{g}^*$, while rigorously handling the noncommutativity of momenta and the topological complications arising from compact subgroups.

Methodology
The authors construct the framework through the following steps:

1. Operator Algebra Construction: Starting from the symplectic structure of the phase space $T^*G$, they define a Poisson bracket where momenta do not commute: $\{p_i, p_j\} = -c_{ij}^k p_k$. They quantize this algebra by promoting coordinates and momenta to operators, adopting a symmetric ordering prescription for momentum products. This choice ensures that the exponential of momentum operators maps directly to exponential operators, satisfying the Baker-Campbell-Hausdorff (BCH) relation: $e^{-i\langle \hat{p}, X \rangle} e^{-i\langle \hat{p}, Y \rangle} = e^{-i\langle \hat{p}, B(X,Y) \rangle}$.

2. Handling Compact Subgroups and Quotients: A critical technical hurdle is that for groups with compact subgroups, the exponential map is not injective. The set of Lie algebra elements exponentiating to the identity, $I = \{X \in \mathfrak{g} \mid \exp(X) = e\}$, is non-trivial.

   • The authors define a "large" position representation on $L^2(\mathfrak{g})$ and a "large" momentum representation on a noncommutative Hilbert space $L^2_\star(\mathfrak{g}^*)$ equipped with a star product (specifically the Gutt star product).
   • They identify that the physical Hilbert space $L^2(G)$ corresponds to the subspace of functions invariant under the action of $I$. They introduce a projection operator $\hat{P}_I$ to enforce this invariance, carefully addressing the infinite normalization factors ($|I| \to \infty$) that arise, analogous to the relation between Fourier transforms on $\mathbb{R}$ and Fourier series on $S^1$.

3. Noncommutative Fourier Transform: They define a unitary intertwiner $\mathcal{F}$ between the large position and momentum representations. The kernel of this transform is the standard plane wave $E(X, p) = e^{-i\langle p, X \rangle}$, but the integration measure includes the Jacobian $J(X)$ of the Haar measure.

   • The transform maps $L^2(\mathfrak{g})$ to $L^2_\star(\mathfrak{g}^*)$.
   • By projecting onto $I$-invariant subspaces, they derive noncommutative Fourier coefficients (mapping $L^2(G)$ to a subspace of $L^2_\star(\mathfrak{g}^*)$) and Fourier series (the inverse map).

4. Star Products and Convolution: The multiplication of functions in momentum space is governed by the star product. The authors demonstrate that convolutions in position space map to star products in momentum space, and vice-versa, establishing a noncommutative analogue of the convolution theorem.

Key Contributions and Results

• Isometry of Hilbert Spaces: The paper proves that the noncommutative Fourier coefficients and series provide an isometry between the Hilbert space of square-integrable functions on the group, $L^2(G)$, and the space of $I$-invariant, star-square-integrable functions on the dual algebra, $\hat{P}_I \cdot L^2_\star(\mathfrak{g}^*)$. This resolves previous ambiguities regarding the well-definedness of scalar products in momentum space for non-Abelian groups.
• Noncommutative Poisson Summation Formula: The authors derive a generalized Poisson summation formula for any compact Lie group. This formula relates the sum of a function over the discrete lattice of logarithms of the identity (in the Lie algebra) to an integral over the dual space involving the Fourier transform and specific differential operators derived from the Jacobian.
  • For $G=U(1)$, this recovers the standard Poisson summation formula.
  • For $G=SU(2)$, they derive a new formula involving integrals over affine planes in $\mathbb{R}^3$, reflecting the non-Abelian nature of the group.
• Kirillov Character Formula: The formalism is applied to the characters of irreducible representations. The authors show that the noncommutative Fourier transform of a character is localized on the corresponding coadjoint orbit. Specifically, for $SU(2)$, the transform of a character with spin $\lambda$ is supported on the sphere of radius $2\lambda + 1$ in $\mathfrak{su}(2)^*$.
• Duflo Ordering: In Section 6, the authors explore an alternative quantization using the Duflo map. They show that while the symmetric ordering leads to "blurred" localization of characters (integration over a family of orbits), the Duflo ordering yields a Fourier coefficient strictly localized on a single coadjoint orbit, directly reproducing the Kirillov character formula. Furthermore, the Duflo-ordered transform naturally leads to the Parseval-Plancherel identity for class functions.

Significance and Claims
The paper claims to provide the necessary "preliminary" framework for computing Wigner functions and path integrals for quantum systems on group manifolds. The authors emphasize that their construction differs from previous approaches in loop quantum gravity (e.g., [23–25]) by:

1. Defining momentum-space wave functions independently of the Fourier transform, then proving the transform is an isometry, rather than defining the space as the image of the transform.
2. Carefully treating the quotients arising from compact subgroups via projection operators, ensuring the scalar product and operator actions are well-defined without relying on ill-defined equivalence classes.
3. Providing a rigorous derivation of the non-Abelian Poisson summation formula, which is identified as a crucial tool for handling the topology of $G$ in path integral calculations (specifically for mixed propagators $\langle p_f | e^{-iHt} | g_i \rangle$).

The work is presented as a mathematical foundation for physicists, aiming to bridge the gap between representation-theoretic approaches (Peter-Weyl) and geometric approaches (phase space path integrals) in the context of Lie-Poisson systems. The authors explicitly state that the results are expected to generalize to infinite-dimensional Lie-Fréchet groups, though this is not rigorously proven in the current text.