Source-linked AI summary
Network Synthesis of Linear Dynamical Quantum Stochastic Systems
H. I. Nurdin, M. R. James, A. C. Doherty
TL;DR
The paper addresses the lack of a comprehensive synthesis theory for general linear dynamical quantum stochastic systems and arbitrary fully quantum controllers. It develops a theorem-based network construction from one-degree-of-freedom open quantum harmonic oscillators, together with quantum-optical implementation schemes. The resulting framework provides systematic physical realizations, illustrated by an explicit two-degree-of-freedom example, while leaving passivity and some unbounded-operator extensions outside its scope.
Problem
A comprehensive method for systematically building arbitrary linear quantum stochastic controllers and general linear dynamical quantum stochastic systems was unavailable, while passivity had not been extensively developed.
Method
The paper develops a synthesis theorem that decomposes general systems into interconnected one-degree-of-freedom open quantum harmonic oscillators and proposes optical implementations using cavities, beam splitters, squeezers, and related elements.
Results
The synthesis theorem establishes an equivalent network realization of the target linear quantum stochastic system, with an explicit two-degree-of-freedom example illustrating its application.
Takeaways & Limitations
The framework provides a systematic route for constructing arbitrarily complex linear dynamical quantum stochastic systems and supports anticipated applications including coherent-feedback controller synthesis.
Takeaways & Limitations
The theory focuses on arbitrary systems without exploiting physical properties such as passivity, and some implementation arguments optimistically assume extensions to unbounded operators.
Abstract
from arXiv · showhide
The purpose of this paper is to develop a synthesis theory for linear dynamical quantum stochastic systems that are encountered in linear quantum optics and in phenomenological models of linear quantum circuits. In particular, such a theory will enable the systematic realization of coherent/fully quantum linear stochastic controllers for quantum control, amongst other potential applications. We show how general linear dynamical quantum stochastic systems can be constructed by assembling an appropriate interconnection of one degree of freedom open quantum harmonic oscillators and, in the quantum optics setting, discuss how such a network of oscillators can be approximately synthesized or implemented in a systematic way from some linear and non-linear quantum optical elements. An example is also provided to illustrate the theory.
1 Background and motivation
The paper addresses the open problem of systematically constructing arbitrary linear quantum stochastic controllers and develops a synthesis framework for general linear dynamical quantum stochastic systems. It motivates this framework through classical network synthesis, quantum Markov modeling, and the physical realization of open oscillator networks.
- Background and motivation: The work targets the open problem of systematically building fully quantum and mixed quantum-classical linear controllers that manipulate quantum signals.Fully quantum controllers, also called coherent-feedback controllers, present new control-design challenges.
- Background and motivation: A new synthesis theorem decomposes arbitrarily complex linear quantum stochastic systems into interconnected one-degree-of-freedom open quantum harmonic oscillators.The theorem provides a systematic construction from basic oscillator building blocks.
- 1.1 Elements of linear electrical network synthesis: The synthesis objective parallels classical electrical network synthesis, which constructs circuits from component-level representations such as state-space descriptions.The paper adapts this systematic inverse-design perspective to quantum stochastic systems.
- 1.1 Elements of linear electrical network synthesis: The paper focuses on arbitrary linear dynamical quantum stochastic systems rather than exploiting properties such as passivity, leaving passivity-related synthesis for future development.Its stated goal is to demonstrate the existence of a physical realization.
- 1.2 Open quantum systems and quantum Markov models: Under the Markov approximation, heat baths behave memorylessly and can be represented as quantum white noise, enabling QSDE descriptions of open-system dynamics.Examples include vacuum noise, squeezed fields, laser fields, and bosonic transmission lines.
- 1.3 Linear dynamical quantum stochastic systems: Linear dynamical quantum stochastic systems model open quantum harmonic oscillators whose canonical position and momentum operators couple linearly to external quantum heat baths.Their dynamics are operator-valued and stochastic because the baths are modeled using quantum noise.
2 Mathematical modeling of linear dynamical quantum stochastic systems
The paper models linear dynamical quantum stochastic systems as open quantum harmonic oscillators driven by bosonic fields, with dynamics represented by operator-valued quantum stochastic equations. Generalized open oscillators additionally include a static passive linear quantum network before the oscillator.
- Linear quantum stochastic systems: An n-degree-of-freedom oscillator uses canonical position and momentum operators satisfying quantum canonical commutation relations and a quadratic Hamiltonian with real symmetric matrix R.The constant term in a quadratic Hamiltonian does not affect oscillator dynamics and can be discarded.
- Quantum stochastic modeling: Markov heat baths are modeled as quantum noise, so their open-system dynamics are naturally described by quantum stochastic differential equations.The bath is idealized as a continuum of harmonic oscillators, producing intrinsically stochastic quantum dynamics.
- Linear quantum stochastic systems: A linear dynamical quantum stochastic system models an open quantum harmonic oscillator whose canonical variables are linearly coupled to external bosonic fields.Its state and output dynamics are linear, but the driving fields are quantum stochastic and the relevant quantities are operator-valued.
- Generalized open oscillators: A generalized open oscillator is the cascade of a static passive linear quantum network and an open oscillator, with the static network applying a unitary scattering transformation to the input fields.This transformation preserves the vacuum noise Ito rules and leaves the overall system dynamics linear.
3 The concatenation and series product of generalized open oscillators and reducible quantum networks
The paper defines concatenation and series products for composing generalized open oscillators, then uses them to describe reducible quantum networks with direct interactions and loop-free connections. These compositions produce another generalized open oscillator while excluding algebraic loops.
- Network composition: Concatenation stacks generalized open oscillators, and for independent oscillators it groups their variables into a larger noninteracting oscillator.The construction permits coincident variables in general, while independence gives the ordinary stacking interpretation.
- Network composition: The series product feeds the output of one generalized open oscillator into the input of another with matching channel counts, yielding another generalized open oscillator.The resulting scattering matrix, coupling operator, and Hamiltonian are determined by the series-product formula.
- Reducible quantum networks: A reducible network combines generalized open oscillators, a direct bilinear interaction Hamiltonian, and specified series connections in which each input and output has at most one connection.The resulting network is again represented as a generalized open oscillator.
- Reducible quantum networks: The framework restricts networks to those without algebraic loops, although more general quantum feedback-network theory can treat such loops.The paper states that this restriction is sufficient for its synthesis theory.
4 Correspondence between system matrices (A, B, C, D) and the parameters S, L, H
The paper establishes a bijective correspondence between the state-space matrices of a generalized open oscillator and its scattering, coupling, and Hamiltonian parameters. This extends the earlier identity-scattering case to arbitrary unitary scattering matrices.
- Physical realizability: The earlier result for S = I gives a bijection between physically realizable system matrices and the open-oscillator parameters K and R.The present formulation extends that correspondence by allowing D = S to be an arbitrary unitary scattering matrix.
- Parameter correspondence: For arbitrary unitary scattering S, transforming the coupling through K′ = S†K reduces the generalized oscillator to an open oscillator with identity scattering.The corresponding system matrices become (A, B, S†C, I).
- Physical realizability: Because the two parameterizations are interchangeable, the paper uses (S, K, R) exclusively for the remainder of its analysis.The generalized-open-oscillator description is therefore both a physical-realization criterion and a convenient synthesis representation.
- Parameter correspondence: Theorem 1 establishes a bijective correspondence between system matrices (A, B, C, D) and generalized-open-oscillator parameters (S, K, R).Given (S, K, R), the system matrices are unique; conversely, D is unitary and determines S.
5 Main synthesis theorem
The main synthesis theorem shows that any n-degree-of-freedom generalized open oscillator can be realized as a reducible network of n one-degree-of-freedom generalized open oscillators. The construction uses cascade connections together with a suitable bilinear direct interaction Hamiltonian, and reduces implementation to synthesizing these components and interactions.
- Theorem 2: Combining two oscillators through cascade and bilinear direct interaction preserves the generalized-open-oscillator form and adds their degrees of freedom while retaining the same numbers of input and output fields.This two-oscillator construction is the recursive step used in the synthesis theorem.
- Theorem 2: Any n-degree-of-freedom generalized open oscillator is equivalent to a cascade of n one-degree-of-freedom generalized open oscillators plus a suitable bilinear direct interaction Hamiltonian.The component scattering and coupling parameters are chosen to reproduce the target scattering matrix S and coupling matrix K, while the direct interaction reproduces the off-diagonal Hamiltonian blocks.
- Theorem 2: The synthesized network uses successive series connections G2 ⊳ G1, G3 ⊳ G2, through Gn ⊳ G(n−1), with component scattering matrices whose product equals the target S.When S is the identity, all component scattering matrices may be chosen as identity and the component couplings as the target coupling blocks.
- Implementation requirements: In principle, arbitrary synthesis requires realizable one-degree-of-freedom open oscillators and a realizable bilinear interaction Hamiltonian Hd.Generalized one-degree-of-freedom oscillators can then be obtained by cascading an open oscillator with a static passive network.
- Relation to electrical synthesis: The construction parallels active state-space synthesis in electrical network theory by treating each oscillator as a noisy quantum integrator and using suitable cascades.The comparison identifies a conceptual analogy rather than an additional synthesis result.
6 Systematic synthesis of linear quantum stochastic systems
This section introduces the quantum-optical components and construction strategy used to synthesize one-degree-of-freedom open oscillators and their interconnections.
- 6 Systematic synthesis of linear quantum stochastic systems: The synthesis procedure combines key quantum-optical components with oscillator construction and approximate bilinear interactions between one-degree-of-freedom open oscillators.The section proceeds from component descriptions to oscillator synthesis and then to direct interaction implementation.
- 6 Systematic synthesis of linear quantum stochastic systems: The proposed implementations use ring-cavity structures with fully and partially reflecting mirrors and appropriately placed linear and nonlinear optical elements.
6.1 Essential quantum optical components
This section develops optical-cavity, parametric-amplifier, and two-mode-squeezing components used as building blocks for linear quantum-optical systems.
- 6.1.1 Optical cavities: Optical cavities confine light through repeated reflection or circulation, while transmitting mirrors allow fields to leak out and introduce losses.
- 6.1.2 Degenerate parametric amplifier: A degenerate parametric amplifier uses a χ(2) crystal and classical pump to amplify a cavity quadrature by converting pump energy into cavity photons.The pump frequency is chosen as ωp = 2ωr, and a rotating-frame description yields a time-invariant model.
- 6.1.3 Two-mode squeezing: Two cavity modes interacting through a pumped χ(2) crystal realize a two-mode squeezing Hamiltonian that simultaneously affects quadratures of both modes.
- 6.1 Essential quantum optical components: The models use a common rotating frame, with all bosonic noises transformed consistently and classical pumps taken at frequency ωp = 2ωr.This setting supports linear time-invariant QSDE models for active optical devices.
6.2 Static linear optical devices and networks
Static linear optical devices implement commutation-preserving field transformations, and arbitrary networks decompose into passive transformations, squeezers, and a final passive transformation.
- 6.2 Static linear optical devices and networks: Quasi-unitary transformations preserve field commutation relations and have quasi-unitary inverses, forming a group under composition.
- 6.2 Static linear optical devices and networks: Passive devices do not mix creation and annihilation operators, and passive networks can be constructed from beam splitters and mirrors.
- 6.2 Static linear optical devices and networks: A phase shifter applies a′ = e^{iθ}a, while a beam splitter combines two input fields through a unitary, energy-conserving transformation.The beam splitter’s mixing angle is its most important parameter, while additional parameters introduce relative or overall phase shifts.
- 6.2 Static linear optical devices and networks: A squeezer reduces one quadrature’s variance at the expense of increasing the other’s, with squeezing parameter s and phase angle θ controlling the transformation.Squeezers can be implemented using a parametric amplifier and beam splitter or a DPA with a transmitting mirror.
- 6.2 Static linear optical devices and networks: Any quasi-unitary network decomposes into two passive networks surrounding independent zero-phase squeezers, so fields are mixed, squeezed, and mixed again.Arbitrary squeezer phase angles can be produced by sandwiching a zero-phase squeezer between two phase shifters.
6.3 Synthesis of one degree of freedom open oscillators
The section realizes arbitrary one-degree-of-freedom open oscillators using cavity Hamiltonians, optical couplings, and auxiliary-mode schemes that approximate general linear coupling operators.
- 6.3.1 Oscillator synthesis: One-degree-of-freedom open oscillators are specified by a real symmetric Hamiltonian matrix R and complex coupling matrix K, both implemented using ring-cavity architectures.
- 6.3.1 Oscillator synthesis: Any real symmetric Hamiltonian matrix R can be realized by choosing the DPA’s complex pump intensity and cavity detuning, with a unique parameter choice for each R.The supplied example gives Δ = 3/2 and ε = −4 − i/2.
- 6.3.2 Linear coupling synthesis: Choosing ε1, ε2, and sufficiently large γ2 allows approximate implementation of arbitrary coefficients α̃ and β̃ in the linear coupling operator.A π-radian phase shifter compensates for the scattering term produced by adiabatic elimination.
- 6.3.3 Alternative coupling synthesis: When α̃ is real and α̃ > |β̃| ≥ 0, an alternative implementation preprocesses and postprocesses squeezed fields around the oscillator.The squeezed input is generated by a squeezer, and the desired output is recovered with the inverse quasi-unitary transformation.
6.4 Engineering the interactions between one-dimensional open quantum harmonic oscillators
The paper implements direct interaction Hamiltonians between one-dimensional oscillators by combining field-mediated optical elements, beam splitters, and χ(2) nonlinear crystals. Under a fast ring-cavity assumption, pairwise interactions can be implemented simultaneously in a network.
- Direct interaction Hamiltonians are decomposed into pairwise terms Hkl = x_k^T C_kl x_l between one-dimensional oscillators.The required total interaction is the sum of direct interactions between oscillator pairs.
- The pairwise Hamiltonian splits into beam-splitter and two-mode-squeezing components in annihilation and creation operators.The beam-splitter component is field-mediated, while the squeezing component uses a χ(2) nonlinear crystal and a classical pump.
- A beam splitter implements the first interaction component with mixing angle Θ = |ε1| and phase Φ = −arg(ε1) + π.The remaining parameters are set as specified for the corresponding optical realization.
- The second interaction component is implemented by a χ(2) nonlinear crystal driven by a pump at frequency 2ωr with effective intensity −2iε2.The two oscillator modes interact through a two-mode squeezing process.
- The overall implementation positions two ring-cavity arms so their circulating beams overlap at beam-splitter and nonlinear-crystal interaction points.This arrangement realizes the pairwise direct interaction while allowing the relevant optical components to act on the cavity modes.
7 Illustrative synthesis example
The illustrative example synthesizes a two-degree-of-freedom open oscillator from two independent one-degree-of-freedom oscillators, a direct interaction Hamiltonian, and optical implementations of each component. The resulting network combines a cascade connection with beam-splitter and nonlinear-crystal interactions.
- Network decomposition: The target system G is constructed as a reducible network from independent oscillators G1 and G2 connected by a direct interaction Hamiltonian Hd12.Because G has identity scattering, Theorem 2 yields the network form G = {I2, G2 ▷ G1} with Hd12 between the subsystems.
- Network decomposition: The physical network combines the cascade G2 ▷ G1 with direct interaction Hd12, as depicted in Figure 5.The example first establishes the block-diagram network before detailing its optical realization.
- Oscillator implementations: G1 is implemented around a ring cavity using a degenerate parametric amplifier, a two-mode squeezer, a beam splitter, and an auxiliary cavity mode.The stated example uses Δ = 5, ε = 1 + i, mirror coupling γ2 = 100, effective pump intensity 10, and beam-splitter mixing angle −10 for the coupling implementation.
- Oscillator implementations: G2 uses the same oscillator construction, with Δ = 2 and ε = 0 for its Hamiltonian and a partially transmitting mirror with κ = 4 for its coupling.Because ε = 0, the implementation of R2 requires no optical crystal or pump beam, only a cavity detuned from the reference frequency.
- Direct interaction implementation: Hd12 is decomposed into a beam-splitter term with mixing angle Θ = −1 and a two-mode-squeezing term implemented in a χ(2) crystal with pump frequency ωp = 2ωr and intensity ε = 4.The two terms are implemented at separate overlap points between the two ring cavities.
- Direct interaction implementation: The final realization passes G1's output Y1(t) into G2 while placing the beam splitter and nonlinear crystal at the cavity-overlap points.Figure 20 presents the physical implementation corresponding to the network block diagram.
8 Conclusions
The paper develops a systematic network theory for synthesizing complex linear dynamical quantum stochastic systems from one-degree-of-freedom open oscillators and proposes quantum-optical implementations. Its appendix also establishes an adiabatic-elimination limit under stated technical assumptions, while noting a bounded-operator caveat for squeezed-noise calculations.
- Conclusions: The paper develops a systematic network theory for synthesizing arbitrarily complex linear dynamical quantum stochastic systems from one-degree-of-freedom open quantum harmonic oscillators.The synthesis includes the required interconnections and interactions among the component oscillators.
- Conclusions: The proposed quantum-optical schemes are intended to support construction of coherent linear quantum stochastic controllers and linear photonic circuits.The stated application scope includes quantum control and quantum information science.
- Adiabatic elimination: For coupled cavity modes, the fast-mode limit yields convergence of the unitary dynamics to a limiting unitary on the slow-mode subspace.The convergence is established for fixed T under the assumptions verified in the appendix.
- Adiabatic elimination: The limiting QSDE coefficients are shown to coincide with the coefficients of the reduced model, completing the adiabatic-elimination proof.The argument verifies the required assumptions and invokes the cited theorem to obtain the convergence result.
- Adiabatic elimination: The adiabatic-elimination construction extends to additional modes that couple to distinct bosonic fields but do not interact with the eliminated mode or one another.Under these conditions, eliminating additional modes contributes only additional terms that exclude the eliminated mode and its associated fields.
- Technical caveat: The squeezed-white-noise calculations rely on an optimistic extension of results derived for bounded Hamiltonian and coupling operators to unbounded oscillator operators.The paper identifies this as a technical caveat rather than resolving the boundedness issue directly.