On the infeasibility of entanglement generation in Gaussian quantum systems via classical control

This paper uses a system theoretic approach to show that classical linear time invariant controllers cannot generate steady state entanglement in a bipartite Gaussian quantum system which is initialized in a Gaussian state. The paper also shows that the use of classical linear controllers cannot generate entanglement in a finite time from a bipartite system initialized in a separable Gaussian state. The approach reveals connections between system theoretic concepts and the well known physical principle that local operations and classical communications cannot generate entangled states starting from separable states.

💡 Research Summary

The paper investigates whether classical linear controllers—both time‑invariant (LTI) and time‑varying (LTV)—can generate entanglement in bipartite Gaussian quantum systems that start in a Gaussian state. The authors model the quantum subsystem as a linear quantum stochastic system (LQSS) described by quantum stochastic differential equations (QSDEs) in the Heisenberg picture. By collecting the canonical position and momentum operators into a real vector x, the dynamics can be written in the familiar state‑space form

dx(t)=A x(t)dt+ B dw(t), dy(t)=C x(t)dt+ D dw(t),

where w(t) represents the quadrature components of the input bosonic fields and satisfies dw dwᵀ = F_w dt with F_w ≥ 0. The matrix A encodes the internal Hamiltonian and coupling to the fields, while B, C, D describe the input‑output relations.

Assuming A is Hurwitz (i.e., the closed‑loop system is asymptotically stable), the steady‑state symmetrized covariance matrix P = lim_{t→∞} ½⟨x(t)x(t)ᵀ + (x(t)x(t)ᵀ)ᵀ⟩ satisfies the real Lyapunov equation

A P + P Aᵀ + B S_w Bᵀ = 0,

with S_w = (F_w + F_w^#)/2. Preservation of the canonical commutation relations yields a second equation

A Θ + Θ Aᵀ – i B T_w Bᵀ = 0,

where Θ = diag(J,…,J) and T_w = (F_w – F_w^#)/(2i). Introducing the complex matrix ˜P = P + i Θ, the two real equations combine into the complex Lyapunov equation

A ˜P + ˜P Aᵀ + B F_w Bᵀ = 0.

Because F_w ≥ 0 and A is Hurwitz, standard results guarantee ˜P ≥ 0, which is precisely the matrix form of the Heisenberg uncertainty principle for Gaussian states.

Entanglement (or separability) of a bipartite Gaussian state is completely characterized by a linear matrix inequality (LMI) due to Simon, Duan, and others: a Gaussian state with covariance P is separable iff

P + i diag(J, –J) ≥ 0.

The authors consider a concrete interconnection: two independent optical cavities (G₁ and G₂) each modeled as an LQSS, linked via a classical controller that receives a homodyne measurement from G₁, processes it, and drives a modulator that injects a control field into G₂. The controller is described by its own state ξ and linear dynamics (A_c, B_c, C_c, D_c). The overall closed‑loop system remains linear, with a block‑diagonal A matrix composed of the plant blocks and the controller block.

The central technical result shows that, for any such classical controller, the steady‑state covariance ˜P of the combined plant‑controller system satisfies ˜P ≥ 0 and, due to the block structure, the plant’s reduced covariance P satisfies the separability LMI. Consequently, the bipartite quantum subsystem is always in a separable (i.e., non‑entangled) state at steady state. Moreover, because the time‑dependent covariance P(t) obeys the same Lyapunov dynamics, the LMI holds for all t ≥ 0; thus no finite‑time entanglement can be created either.

These findings are a direct system‑theoretic manifestation of the well‑known LOCC (Local Operations and Classical Communication) principle: classical processing of measurement results and classical feedback cannot generate entanglement from an initially separable state. The paper therefore establishes a rigorous no‑go theorem for entanglement generation using only classical linear control in Gaussian quantum optics.

Implications are significant for quantum control engineering. Any scheme that relies solely on classical measurement‑based feedback (e.g., homodyne detection plus linear electronic controllers) cannot be used to produce or enhance entanglement; one must resort to genuinely quantum resources such as direct quantum‑quantum interactions, non‑linear quantum Hamiltonians, or measurement‑based quantum feedback that retains quantum coherence. Additionally, the work illustrates how classical control theory tools—Lyapunov equations, stability analysis, and LMIs—can be leveraged to prove fundamental quantum information limits, opening avenues for further cross‑disciplinary investigations into quantum‑classical interface constraints.