Rigorous s-ordered phase-space representations: unbounded operators, ultradistributional weight functions, and the singularity boundary x(F) of the Cahill–Glauber ordering continuum

Cahill and Glauber's 1969 paper on ordered expansions in boson amplitude operators introduced the s-parametrized ordering convention, in which normal, symmetric (Weyl), and antinormal ordering correspond to s=+1, 0, −1 of a continuous order parameter, together with the corresponding ordered power series and integral representations of operators. The paper itself acknowledged three unresolved issues, confirmed in the primary text: (1) no rigorous formulation of the integral representations for unbounded operators; (2) weight functions that, for some s, are more singular than tempered distributions and would require the representation structure to be 'changed in a major way' to be interpreted as ultradistributions; (3) an undetermined operator-dependent boundary x(F) and undetermined singularity locations s0 in the complex s-plane, which can lie arbitrarily close to the line Re s = 0 even for trace-class operators. The paper additionally remarked that the new intermediate orderings were introduced without any claim of 'direct physical significance'. Later work has partially addressed item (2) and supplied a physical interpretation of the s ≤ 0 half of the ordering continuum: the singularity of the Glauber–Sudarshan P function was characterized within distribution theory with a maximally singular bound, operational regularization schemes were developed, and for s ≤ 0 the s-ordered distributions were identified as Wigner functions of states degraded by Gaussian noise channels. Items (1) and (3) remain open, as does the ultradistributional restructuring required for all s. The surviving open core is a rigorous generalized-function and operator-theoretic foundation of the s-ordered representations, together with the classification of the singularity boundary x(F).

Background

The source paper concerns the quantum theory of a single bosonic mode, described by annihilation and creation operators $a$ and $a^\dagger$ satisfying $[a, a^\dagger] = 1$ . Because $a$ and $a^\dagger$ do not commute, an operator expression such as a product of powers of $a$ and $a^\dagger$ is ambiguous until an ordering convention is fixed: normal ordering places all $a^\dagger$ to the left, antinormal ordering places all $a$ to the right, and symmetric ordering averages over all orderings. Phase-space methods relate operators to c-number functions, and the three orderings produce the familiar quasiprobabilities (functions that behave like probability densities but may be negative or singular): the Glauber–Sudarshan P function (normal side), the Wigner function (symmetric ordering, realized by the Weyl correspondence), and the Husimi Q function. Cahill and Glauber introduced a one-parameter family of orderings indexed by a complex order parameter $s$ , with normal, symmetric, and antinormal ordering at $s = +1, 0, -1$ , and showed that for bounded operators the coefficients of the $s$ -ordered power series are finite for $s \geq 0$ and the series converge for $s \geq 1/2$ ; in the source's terminology 'bounded' means of finite Hilbert–Schmidt norm $\|F\|_2 = [\operatorname{Tr}(F^\dagger F)]^{1/2}$ , while 'finite' means of finite operator (least upper bound) norm — all bounded operators are finite, and the unitary displacement operators are the standard examples of operators that are finite but unbounded. They defined kernels $T(\alpha, s)$ — complex Fourier transforms of the displacement operators $D(\xi) = \exp(\xi a^\dagger - \xi^* a)$ — which reduce at $s = -1$ to the coherent-state projectors $|\alpha\rangle\langle\alpha|$ , and the integral representation (their Eq. (6.11)) $F = \int f(\alpha, -s) T(\alpha, s) \pi^{-1} d^2\alpha$ with weight function $f(\alpha, -s) = \operatorname{Tr}[F T(\alpha, -s)]$ . Note that the kernel argument and the weight-function argument here carry opposite signs ( $T(\alpha, s)$ is paired with $f(\alpha, -s)$ ), so the weight function $f(\alpha, -s)$ of Eq. (6.11) and the $f(\alpha, s)$ of part (3) are the same one-parameter family read at the negated argument; it is the weight-function argument that is the ordering parameter of the family, normal, symmetric, and antinormal ordering lying at $s = +1, 0, -1$ . The weight functions at different $s$ are related by Gaussian convolution (a heat-type equation in $s$ ), so increasing $s$ de-smooths the weight function toward the P function. For the family $F(\lambda) = \lambda^N$ with $N = a^\dagger a$ the photon-number operator, the paper computed closed-form weight functions that are analytic in $s$ up to a first singularity at $s_0 = (1+\lambda)/(1-\lambda)$ , and it defined $x(F)$ as the boundary of the region in the $s$ -plane in which the weight function is bounded, smooth, and square-integrable; for trace-class operators (operators of finite trace norm, the class to which density operators belong) this boundary can lie arbitrarily close to the line $\operatorname{Re} s = 0$ . A tempered distribution is a continuous linear functional on the Schwartz space of smooth, rapidly decreasing test functions (it includes the Dirac delta and its derivatives); the paper noted, following Gel'fand and Shilov's generalized-function theory, that beyond tempered distributions lies the larger class of ultradistributions (the dual space $Z'$ ), and that for all $s$ and all bounded operators the weight function $f(\alpha, -s)$ can be interpreted as an ultradistribution, provided the structure of the representation (6.11) is 'changed in a major way'. The source paper left three issues open, explicitly acknowledged by the authors — the absence of a careful formulation for unbounded operators ('we shall not in this paper attempt a careful formulation of the representation of unbounded operators'); the ultradistributional weight functions requiring that 'the structure of the representation (6.11) must be changed in a major way'; and the undetermined general behavior of the line $\operatorname{Re} s = x(F)$ and of the singularities $s_0$ .

Problem Statement

For the Cahill–Glauber s-parametrized ordering of boson operators, establish the missing mathematical foundation of the integral representations, in the three parts left open by the source paper and not closed by subsequent literature (the three parts are jointly required for the foundation sought here, and the admissible operator classes in part (1) are to be fixed and justified explicitly as part of the solution): (1) Unbounded operators — formulate the displacement-operator and $T(\alpha, s)$ -based integral representations rigorously for physically important unbounded operators (the source notes that even operators that are 'finite but unbounded' in its terminology — of finite operator norm but not of finite Hilbert–Schmidt norm, as are the unitary displacement operators $D(\xi)$ — fall outside the existing theory, their weight functions already being singular), specifying the admissible operator classes, the function spaces for the weight functions, and the sense in which the expansions represent the operator; this includes justifying or repairing the steps in the original derivations that apply bounded-operator representations to the kernels $T(\alpha, s)$ , which are themselves unbounded for $\operatorname{Re} s > 0$ . (2) Singular weight functions — construct the generalized-function (ultradistributional) framework and the corresponding restructuring of the integral representation (6.11) such that the restructured representation is valid for all bounded operators and all values of the order parameter $s$ , with the weight functions interpreted as ultradistributions: the source showed that for $\operatorname{Re} s > 0$ the kernels $T(\alpha, s)$ are undercomplete (for no value of $\operatorname{Re} s > 0$ do they afford a basis for the representation (6.11) of an arbitrary bounded operator) even when the whole class of tempered distributions is admitted as weight functions, and stated that the weight functions $f(\alpha,-s)$ can be interpreted as ultradistributions, for all $s$ and all bounded operators, only if 'the structure of the representation (6.11) must be changed in a major way'; determine what that restructured representation is and reconcile it with the now-known maximal degree of singularity of the P function and the analytic test-function class identified in later distribution-theoretic work. (3) Singularity structure in the s-plane — determine, for general operator classes, the boundary $x(F)$ (the maximal $\operatorname{Re} s$ for which the weight function $f(\alpha, s)$ is bounded, smooth, and square-integrable) and the locations and nature of the first singularities $s_0$ of $f(\alpha, s)$ as an analytic function of $s$ , given the known facts that these quantities vary widely from operator to operator, that $x(F)$ can be arbitrarily close to the line $\operatorname{Re} s = 0$ within the trace class, and that the closed-form family $F(\lambda) = \lambda^N$ realizes $s_0 = (1+\lambda)/(1-\lambda)$ throughout the half-plane $\operatorname{Re} s > 0$ . The required form of this determination is an explicit criterion or classification that decides $x(F)$ and the location and nature of the first singularity $s_0$ for a general admissible operator $F$ — or a demonstration that no such uniform criterion exists; a bound valid for all operators, or the analysis of the single closed-form family $F(\lambda)$ , does not by itself suffice for this part.

References

K. E. Cahill and R. J. Glauber, Ordered Expansions in Boson Amplitude Operators, Physical Review 177, 1857–1881 (1969), doi:10.1103/PhysRev.177.1857, https://doi.org/10.1103/PhysRev.177.1857.
K. E. Cahill and R. J. Glauber, Density Operators and Quasiprobability Distributions, Physical Review 177, 1882–1902 (1969), doi:10.1103/PhysRev.177.1882, https://doi.org/10.1103/PhysRev.177.1882.
J. Sperling, Characterizing maximally singular phase-space distributions, Physical Review A 94, 013814 (2016), doi:10.1103/PhysRevA.94.013814, https://doi.org/10.1103/PhysRevA.94.013814.
Y. Zhang and S. Luo, From the Wigner function to the s-ordered phase-space distribution via a Gaussian noise channel, Theoretical and Mathematical Physics 210, 425–441 (2022), doi:10.1134/S0040577922030126, https://doi.org/10.1134/S0040577922030126.
B. Kühn and W. Vogel, Preparation of arbitrary quantum states with regular P functions, Physical Review A 98, 053807 (2018), doi:10.1103/PhysRevA.98.053807, https://doi.org/10.1103/PhysRevA.98.053807.
M. Zartab, E. Shojaee, and S. Rahimi-Keshari, Filter functions for the Glauber-Sudarshan P-function regularization, Physical Review A 107, 053706 (2023), doi:10.1103/PhysRevA.107.053706, https://doi.org/10.1103/PhysRevA.107.053706.
J. Sperling and W. Vogel, Quasiprobability distributions for quantum-optical coherence and beyond, Physica Scripta 95, 034007 (2020), doi:10.1088/1402-4896/ab5501, https://doi.org/10.1088/1402-4896/ab5501.
B. Kühn and W. Vogel, Unbalanced Homodyne Correlation Measurements, Physical Review Letters 116, 163603 (2016), doi:10.1103/PhysRevLett.116.163603, https://doi.org/10.1103/PhysRevL.116.163603.