Source-linked AI summary
The Series Product and Its Application to Quantum Feedforward and Feedback Networks
J. Gough, M. R. James
TL;DR
Quantum-network analysis needs general algebraic descriptions of component connections. This paper introduces series and concatenation products, uses them to model feedforward, feedback, and reducible networks, and illustrates the framework with quantum-control examples.
Problem
The paper addresses the need for simple, general methods to describe series connections in quantum networks, including more general interfaces.
Method
The authors introduce a parametric representation for open quantum systems together with concatenation and series products for assembling systems and modeling field-mediated connections.
Results
The products describe feedforward, feedback, reducible, and other quantum networks, with examples from the quantum-control literature analyzed using the framework.
Takeaways & Limitations
The framework provides a simple, transparent way to express quantum feedback-control and quantum-filtering examples.
Takeaways & Limitations
The paper restricts its treatment to canonical quantum signals and does not pursue modifications required for non-canonical input correlations.
Abstract
from arXiv · showhide
The purpose of this paper is to present simple and general algebraic methods for describing series connections in quantum networks. These methods build on and generalize existing methods for series (or cascade) connections by allowing for more general interfaces, and by introducing an efficient algebraic tool, the series product. We also introduce another product, which we call the concatenation product, that is useful for assembling and representing systems without necessarily having connections. We show how the concatenation and series products can be used to describe feedforward and feedback networks. A selection of examples from the quantum control literature are analyzed to illustrate the utility of our network modeling methodology.
1 Introduction
The paper develops algebraic methods for quantum-network series connections, extending prior approaches with more general interfaces and a series product. It also introduces concatenation and applies both products to feedforward, feedback, and quantum-control examples.
- Motivation: The paper presents simple, general algebraic methods for describing series connections in quantum networks.
- Contributions: The series product extends earlier quantum series-connection results by accommodating more general interfaces.
- Contributions: The concatenation product assembles systems without necessarily connecting them, while the series product represents field-mediated connections.
- Scope: The framework supports modeling open quantum physical systems and networks with boson-field interconnects such as optical beams or phonon vibrations.
- Applications: The authors use the products to express quantum feedback-control and quantum-filtering examples in a simple, transparent way.
- Organization: The paper develops its theory through classical analogs, quantum-component examples, open quantum stochastic models, network results, and technical appendices.
2 Classical Linear Systems
The classical treatment recasts concatenation and series connections as algebraic operations on system parameters and transfer functions. Concatenation keeps systems assembled but unconnected, whereas the series product models cascaded signal flow.
- Overview: The paper introduces classical algebraic analogs to clarify the quantum concatenation and series products.
- Concatenation product: Concatenation assembles two systems without connections and yields the block-diagonal transfer function G(s) = diag{G1(s), G2(s)}.
- Series product: The series connection sets u2 = y1, requiring dim u2 = dim y1, and is represented by the transfer-function product G(s) = G2(s)G1(s).
- Series product: The series product describes a cascade connection fundamental to feedforward and feedback control.
- Algebraic representation: Both products are defined using system parameters, including state-space parameters or transfer-function matrices.
3 Example Components and Connections
This section introduces quantum-mechanical foundations and models optical cavities as open systems coupled to external quantum fields. It also describes cavity input-output behavior and the role of beamsplitter-style field representations.
- Quantum mechanics: Quantum observables are self-adjoint operators on a Hilbert space, while state vectors summarize system states and determine observable expectations.
- Quantum mechanics: The Heisenberg picture evolves observables through unitary dynamics and is used because it relates closely to classical control and probability models.
- Quantum harmonic oscillator: A harmonic oscillator has discrete energy levels, with annihilation and creation operators satisfying [a, a∗] = 1.
- Optical cavities: An optical cavity traps a harmonic-oscillator mode between partially transmitting and perfectly reflecting mirrors, allowing interaction with an external free field.
- Optical cavities: The cavity's input-output structure treats one traveling field component as input and the other as output carrying information after interaction with the cavity mode.
- Open cavity models: The cavity-free-field model uses a Hamiltonian for internal energy and field coupling, then represents the open system with a unitary QSDE and output-field equation.
4 Open Quantum Stochastic Models
This section specifies open quantum systems by scattering, coupling, and Hamiltonian parameters and derives their unitary, Heisenberg, output-field, and master-equation descriptions. The framework assumes canonical quantum inputs and supports conditional quantum filtering for monitored outputs.
- Model parameters: An open quantum model is parameterized as G = (S, L, H), where S is scattering, L is coupling, and H is the self-energy Hamiltonian.
- Model parameters: The scattering matrix is unitary, the coupling vector describes energy exchange with fields, and the Hamiltonian describes the system's internal energy.
- Canonical inputs: Canonical quantum inputs obey specified non-vanishing second-order Itō products, with all differentials interpreted in the Itō sense.
- System dynamics: The Schrödinger equation with V(0) = I determines unitary system motion, while Heisenberg evolution applies the resulting dynamics to initial-space operators.
- Output fields: Output fields satisfy quantum stochastic differential equations and retain canonical quantum Itō products under the open-system dynamics.
- Master equations and filtering: The parameters G = (S, L, H) determine master equations and quantum filters, while S additionally specifies input-channel architecture.
5 The Concatenation and Series Products and their Application to Quantum Networks
The paper develops concatenation and series products for representing quantum components, their connections, and reducible networks. These products support feedback, cascade, and network construction, including equivalent-system transformations and applications to quantum control examples.
- 5.1 Definitions: The concatenation and series products are defined as the main algebraic tools for assembling quantum systems and modeling their connections.Concatenation combines systems without field-channel interconnections, while the series product represents feeding one system’s output into another’s input.
- 5.2 Feedback: The series product gives the parameters of the feedback system formed when the output of the first subsystem is fed into the input of the second.This is stated as the Principle of Series Connections in Theorem 5.5.
- 5.3 Cascade: Series connections also apply to cascades of independent systems, although the series product is not generally commutative and swapping components can require modifying one component.Theorem 5.6 characterizes when two cascaded systems are parametrically equivalent after such a modification.
- 5.3 Cascade: A scattering matrix can be moved from a system’s input to its output by transforming the coupling vector to S†L while retaining the Hamiltonian H.The identity expresses the same parameters through two alternative series-product factorizations.
- 5.4 Reducible Networks: Reducible networks combine maximal series chains and uninvolved components through concatenation, while direct interactions are specified by a connection Hamiltonian.The framework permits both direct interactions and indirect interactions through field interconnects, subject to matching dimensions and at-most-one connections per input or output.
6 Examples
The paper applies its network theory to all-optical, direct measurement, and realistic detection feedback schemes, representing them as reducible networks and deriving closed-loop dynamics. The examples include photon counting, homodyne detection, and filtering with classical noise.
- The examples are drawn from quantum-control literature and are represented using reducible networks.The section presents several literature-based feedback and detection examples through the paper’s network methodology.
- 6.1 All-Optical Feedback: All-optical cavity feedback is modeled as a light path between two partially transmitting mirrors with an intervening phase shift.The cavity example treats the phase shift as a separate system component and forms equivalent network representations.
- 6.1 All-Optical Feedback: The closed-loop cavity system is obtained by applying the series-connection formulas or by moving the phase component using equivalent-component theory.Both network representations yield the same closed-loop description and Heisenberg cavity-mode equation.
- 6.2 Direct Measurement Feedback: Direct measurement feedback is formulated for photon counting and homodyne detection, with proportional feedback represented through a coupling operator and gain.The gain can be absorbed into the self-adjoint control operator, allowing the subsequent discussion to assume unit gain.
- 6.2 Direct Measurement Feedback: The photon-counting and homodyne constructions produce Heisenberg equations agreeing with the corresponding results of Wiseman.The photon-counting case uses a coupling that selects the photon-number observable, while the homodyne case selects a field quadrature.
- 6.3 Realistic Detection: Realistic homodyne detection is modeled by appending a classical filter and additive noise to ideal homodyne measurement, then representing the arrangement as a reducible network.For linear Gaussian quantum and classical systems, the resulting complete filter reduces to a Kalman filter for estimating quantum variables.
7 Conclusion
The paper concludes that its parametric, concatenation, and series-product tools model quantum networks and reducible networks, while noting a scope boundary in the Holevo-to-Stratonovich correspondence. Future work targets network-theory development and quantum-technology applications.
- The paper presents a parametric representation together with concatenation and series products as algebraic tools for modeling quantum networks.Concatenation assembles components without necessarily connecting them, while the series product combines systems through field-mediated connections.
- The theory models reducible quantum networks and illustrates its use through examples from the quantum-control literature.The examples serve as demonstrations of the network-modeling methodology.
- Future work concerns further development of the network theory and applications to control-engineering tools and quantum technology.The conclusion cites quantum-technology applications as a future direction.
- The Holevo generating coefficients correspond to the parameters G = (S, L, H), but the implicit-explicit formalism coincides with the Stratonovich-Itō correspondence only when H11 = 0.This states the paper’s explicit scope condition for that correspondence.
B Proof of Theorem 5.5
The proof models quantum cascade connections through discrete qubit interactions and takes a continuous-time limit. The limiting QSDE coefficients are identified with the series product of the component parameters.
- Discrete model: The input field is discretized as a beam of independently prepared qubits interacting with plants at discrete times.For two cascaded plants, the qubits are separated by a time of flight of exactly τ seconds.
- Discrete model: Discrete-time dynamics are constructed from repeated unitary kicks, with separate evolutions for single-system and cascaded interactions.The proof expands the kick exponentials and iterates the resulting expressions.
- Continuous-time limit: As τ →0+, the discrete single-system process converges weakly in matrix elements to the corresponding QSDE solution.The coefficients of the unitary QSDE are related to the first system’s Hamiltonian and parameters.
- Continuous-time limit: Under the same convergence mode, the cascaded dynamics converge to a limiting QSDE.The discrete processes used in the proof approximate the fundamental quantum stochastic processes as τ tends to zero.
- Series-product identification: The limiting cascade coefficients equal those generated by the series product G2 ⊳ G1, establishing G2←1 ≡ G2 ⊳ G1.The result extends directly to multidimensional noise.
C Proof of Theorem 5.6
The proof establishes parameter equivalence for cascade systems and then embeds a classical stochastic system as a commutative subsystem of a quantum network. The resulting homodyne output and filter recover the classical model and nonlinear filtering equation.
- Cascade equivalence: Cascade systems are equivalent when their component parameters satisfy the stated matching conditions.The proof compares the scattering, coupling, and Hamiltonian terms to solve for the primed parameters.
- Classical embedding: The classical system has state x(t) ∈ R^n, scalar output y(t), smooth vector fields, and independent standard Wiener processes.The construction seeks a quantum network representation of this classical system.
- Classical embedding: The classical block diagram and its network representation are used to model the system as a commutative subsystem of a larger quantum system.The classical noises are represented as real quadratures of quantum noises.
- Classical embedding: The embedded classical dynamics reside in the position observable and are unaffected by the second input field.Only the real quadrature of the first input field affects these dynamics.
- Filtering correspondence: The homodyne detector output agrees with the classical output, and the unnormalized quantum filter is the Duncan-Mortensen-Zakai equation.Thus the quantum network representation reproduces the classical nonlinear filtering model.