Double Single-Sided Cavity System

• Double single-sided cavity systems are composite architectures in which each cavity features one input–output port and a perfectly reflecting boundary, allowing controlled interference and coupling.
• They leverage tailored boundary conditions and mediators like mechanical resonators or two-level systems to achieve functionalities such as squeezing, entanglement transfer, and nonreciprocal photon blockade.
• This modular design separates roles such as strong intracavity field build-up from low-loss readout, optimizing performance in varied regimes and applications.

Searching arXiv for recent and foundational papers on double single-sided cavity systems and closely related architectures. A double single-sided cavity system is a composite cavity-QED or optomechanical architecture built from two single-sided optical cavities, each having one coupling channel that serves as the input–output port while the other boundary is ideally perfectly reflecting. Across the literature, this label does not denote a single universal Hamiltonian; rather, it denotes a family of two-cavity platforms in which single-sided boundary conditions are central to the dynamics, control, and readout. In optomechanics, the configuration appears both as two directly coupled cavities with one cavity acting as an auxiliary channel (Wang et al., 2016), and as two independent single-sided cavities coupled indirectly through a common mechanical resonator (Huan et al., 2015). Closely related variants include two single-sided cavities each containing a Bose–Einstein condensate and driven by entangled light (Kumar et al., 2010), nonreciprocally coupled optomechanical cavities for photon blockade (Wang et al., 2020), and two perpendicular single-sided cavities coupled to a two-level system for photonic quantum gates (Li et al., 30 Jul 2025). The common structural feature is that the input–output physics is organized around one external channel per cavity, which makes interference, scattering asymmetry, conditional phase response, and port-resolved measurement especially transparent.

1. Defining architectures and physical realizations

The most direct realization is the two-cavity optomechanical configuration studied in “Steady-state mechanical squeezing in a double-cavity optomechanical system” (Wang et al., 2016). There, the system comprises two coupled single-mode optical cavities with intercavity photon-hopping rate $J$. Cavity 1 is strongly driven and directly optomechanically coupled to a mechanical resonator of frequency $\omega_m$, whereas cavity 2 is a high-$Q$ auxiliary cavity without direct optomechanical coupling. Each cavity is treated as single-sided, with a single decay rate $\kappa_1$ or $\kappa_2$ and corresponding input noise operators $a_{1in}$ and $a_{2in}$ (Wang et al., 2016). This arrangement is explicitly designed so that a highly dissipative primary cavity can build up a large intracavity field while a low-loss auxiliary cavity shapes the effective optical response seen by the mechanics.

A different optomechanical realization appears in “Dynamic entanglement transfer in a double-cavity optomechanical system” (Huan et al., 2015). In that case, two independent single-sided optical Fabry–Perot cavities are formed on the two sides of a reflective mechanical element. The left and right cavity modes do not tunnel directly into one another; instead, both couple through radiation pressure to the same mechanical coordinate, so the mechanical resonator mediates an indirect photon–photon interaction. Here again, “single-sided” means each cavity has a single input-output coupling mirror through which light both enters and exits, with the other mirror ideally perfectly reflecting (Huan et al., 2015).

The same single-sided boundary-condition logic extends beyond conventional optomechanics. In “Entangling two Bose Einstein condensates in a double cavity system” (Kumar et al., 2010), two physically separated cavities are each driven and read out from the same mirror, and a Faraday isolator enforces unidirectional coupling of NOPA-generated entangled light into the cavities. In “General quantum computation on photons assisted with double single-sided cavity system” (Li et al., 30 Jul 2025), two single-sided optical cavities cross mutually perpendicularly and both couple to a common two-level system. In that setting, the two single-sided ports become the natural logical scattering channels for reflection, transmission, and conditional routing.

This diversity suggests that “double single-sided cavity system” is best understood as a boundary-condition-defined two-cavity platform rather than a single model. A plausible implication is that the phrase is most useful when the one-port-per-cavity structure directly determines the accessible interference pathways and measurable observables.

2. Single-sided input–output structure

The single-sided condition is operational rather than merely geometric. In the optomechanical squeezing model, the standard input–output relations are $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$ and $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$ (Wang et al., 2016). These relations are consistent with the Markovian Heisenberg–Langevin equations used for the intracavity fields and underlie detection and noise spectra.

In the mechanically mediated double-cavity system, the single-sided relations are written as $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$ for $\omega_m$0 (Huan et al., 2015). The optical inputs are taken to be vacuum, with delta-correlated fluctuations, while the mechanical bath is thermal and Markovian in the high-$\omega_m$1 limit. This permits construction of the full covariance-matrix evolution for the mechanical and optical quadratures (Huan et al., 2015).

In the BEC-cavity configuration, the corresponding relations are $\omega_m$2 (Kumar et al., 2010). The cavities are explicitly treated as single-sided because the same mirror provides both the driving interface and the collection port. The NOPA outputs feed the two cavities unidirectionally, so the port geometry is integral to the entanglement-transfer mechanism (Kumar et al., 2010).

For the crossed-cavity TLS architecture, the Heisenberg–Langevin form is

$\omega_m$3

$\omega_m$4

for each single-sided cavity $\omega_m$5 (Li et al., 30 Jul 2025). In that model, if the TLS is decoupled, each cavity behaves as an independent reflective single-sided resonator; when the TLS couples to both cavities, the atom–cavity hybridization opens a transmission path between the ports (Li et al., 30 Jul 2025).

A recurring misconception is that two cavities automatically imply symmetric transmission through the device. The literature here shows otherwise. In a single-sided implementation, each cavity’s natural external response is reflection-dominated unless an internal mediator—mechanical, atomic, or intercavity—creates an alternate channel (Wang et al., 2016, Li et al., 30 Jul 2025).

3. Core Hamiltonian classes

Because the term covers multiple physical settings, the Hamiltonian depends on the mediator that couples the two single-sided cavities.

In the directly coupled optomechanical case (Wang et al., 2016), the total Hamiltonian is

$\omega_m$6

with

$\omega_m$7

$\omega_m$8

$\omega_m$9

This model combines direct photon hopping, radiation-pressure coupling localized to cavity 1, coherent pumping, and mechanical Duffing nonlinearity (Wang et al., 2016).

In the shared-mechanics architecture (Huan et al., 2015), the lab-frame Hamiltonian is

$Q$0

$Q$1

There is no direct optical tunneling term; the shared resonator produces the effective intercavity coupling dynamically through its susceptibility $Q$2 (Huan et al., 2015).

In the nonreciprocal photon-blockade system (Wang et al., 2020), the optical cavities are linked by asymmetric hopping: $Q$3 Under the polaron transformation and for $Q$4, this reduces to an effective Kerr model with $Q$5 (Wang et al., 2020).

In the crossed-cavity TLS model (Li et al., 30 Jul 2025), the full Hamiltonian includes two cavity modes, two continua, and a two-level system: $Q$6

$Q$7

$Q$8

This Hamiltonian is not optomechanical, but it preserves the defining two-single-sided-cavity structure (Li et al., 30 Jul 2025).

4. Interference, effective models, and regime engineering

The principal conceptual utility of the double single-sided cavity system lies in its ability to engineer effective interactions that a single cavity cannot realize under the same dissipation constraints.

In the squeezing protocol of (Wang et al., 2016), cavity 1 is intentionally very lossy, with $Q$9, so that strong driving can generate a large intracavity amplitude and hence a large linearized coupling $\kappa_1$0. Cavity 2 is a narrow auxiliary channel with $\kappa_1$1. By adiabatically eliminating cavity 1 in the regime $\kappa_1$2, the reduced dynamics becomes

$\kappa_1$3

$\kappa_1$4

with effective parameters

$\kappa_1$5

$\kappa_1$6

The paper interprets these as interference-modified parameters that suppress Stokes heating and preserve anti-Stokes cooling even when the original optomechanical cavity is deep in the unresolved-sideband regime (Wang et al., 2016).

In the shared-resonator architecture (Huan et al., 2015), elimination of the mechanics to leading order gives an effective photon–photon coupling

$\kappa_1$7

This coupling is frequency dependent and inherits the thermal and dissipative structure of the mechanical bus (Huan et al., 2015). That feature distinguishes it from direct optical tunneling and is central to the observed delay, saturation, and death-and-revival phenomena in entanglement transfer.

In the nonreciprocal blockade setting (Wang et al., 2020), the interference is not between cavity supermodes generated by symmetric hopping but between two-photon excitation pathways weighted by $\kappa_1$8 and $\kappa_1$9. The unconventional photon blockade condition is the cancellation of the two-photon amplitude $\kappa_2$0,

$\kappa_2$1

The paper emphasizes that the behavior of blockade under nonreciprocity differs qualitatively between CPB and UPB: UPB has a sharp optimum in directionality, whereas CPB improves monotonically as $\kappa_2$2 increases (Wang et al., 2020).

These examples indicate that the “double” character is not merely multiplicative; it permits an engineered effective response that can be narrower, more directional, or more interference-sensitive than either cavity alone. This suggests that the two-cavity, one-port-per-cavity structure is especially useful when one wants to separate field buildup from readout bandwidth or separate strong local nonlinearity from low-loss transport.

5. Representative phenomena

Phenomenon	Mechanism	Representative paper
Steady-state mechanical squeezing	Coherent auxiliary-cavity interference with effective cooling in transformed frame	(Wang et al., 2016)
Dynamic intercavity entanglement transfer	Shared mechanical mediator induces indirect photon–photon coupling	(Huan et al., 2015)
BEC–BEC EPR entanglement	NOPA-generated optical entanglement mapped into collective density modes	(Kumar et al., 2010)
Photon blockade	Kerr nonlinearity plus nonreciprocal interference pathways	(Wang et al., 2020)
Single/double Fano resonances	Probe interference via coupled-cavity supermodes and optomechanics	(Prakash et al., 2019)
Photonic controlled-phase gates	TLS-mediated conditional reflection and transmission between ports	(Li et al., 30 Jul 2025)

For mechanical squeezing, the central observable is the steady-state variance of the mechanical displacement quadrature $\kappa_2$3,

$\kappa_2$4

with squeezing parameter

$\kappa_2$5

The optimal cooling condition is $\kappa_2$6 in the transformed frame, and the paper states that the maximum of the squeezing parameter occurs near $\kappa_2$7 (Wang et al., 2016).

For entanglement transfer, the covariance matrix $\kappa_2$8 obeys

$\kappa_2$9

and the logarithmic negativity

$a_{1in}$0

is used to quantify bipartite entanglement (Huan et al., 2015). The paper reports two distinct dynamical regimes: a saturation regime and a death-and-revival regime, with a finite “time lapse” before intercavity entanglement appears (Huan et al., 2015).

For the BEC implementation, the paper uses normalized EPR variances $a_{1in}$1 and $a_{1in}$2 and takes entanglement to occur when $a_{1in}$3 (Kumar et al., 2010). The transfer is optimized by the algebraic condition

$a_{1in}$4

with NOPA operation near but below threshold, $a_{1in}$5 (Kumar et al., 2010).

In the Fano-resonance system, changing the photon-hopping rate $a_{1in}$6 of the middle mirror switches the reflected probe from single-Fano to double-Fano line shapes (Prakash et al., 2019). The paper states that the first spectral line is stronger in the multi-Fano case and that the steady-state displacement of the middle mirror strongly influences the double-Fano structure (Prakash et al., 2019).

In the quantum-gate architecture, the two-port scattering matrix is

$a_{1in}$7

with explicit $a_{1in}$8 and $a_{1in}$9 determined by $a_{2in}$0, $a_{2in}$1, and the detunings (Li et al., 30 Jul 2025). Under resonant, symmetric conditions, the paper gives

$a_{2in}$2

These relations allow the TLS to function as a conditional switch between the two single-sided ports (Li et al., 30 Jul 2025).

6. Parameter regimes, stability, and experimental considerations

A defining technical theme in this literature is that the double single-sided configuration is often introduced to relax a standard single-cavity constraint.

In (Wang et al., 2016), the representative unresolved-sideband regime uses $a_{2in}$3, $a_{2in}$4, $a_{2in}$5, $a_{2in}$6, $a_{2in}$7, and $a_{2in}$8, together with $a_{2in}$9 MHz and optical-scale cavity frequency $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$0 THz. With $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$1 mW, the paper reports $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$2, $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$3, and hence $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$4 (Wang et al., 2016). The point of this regime is precisely that large $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$5 is obtained despite $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$6.

That same paper states that linearized stability is ensured when the drift matrix is Hurwitz, although no explicit Routh–Hurwitz criterion is given; stability is checked numerically. It also notes that $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$7 is real only if $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$8, so excessive $a_{1,out} = a_{1,in} + \sqrt{\kappa_1} a_1$9 risks parametric instability (Wang et al., 2016).

In the dynamic entanglement-transfer setting (Huan et al., 2015), resolved-sideband operation is favorable. The symmetric example quoted in the paper uses mechanical $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$0, $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$1 MHz, effective mass $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$2 ng, cavity length $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$3 mm, finesse $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$4, wavelength $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$5 nm, and powers $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$6W. Using $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$7, one gets $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$8, so $a_{2,out} = a_{2,in} + \sqrt{\kappa_2} a_2$9 (Huan et al., 2015). Lower finesse and higher power produce shorter delays and more pronounced collapse–revival behavior (Huan et al., 2015).

In the photon-blockade proposal (Wang et al., 2020), the weak-drive regime assumes $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$0. Representative parameters are $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$1 MHz, $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$2 MHz, $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$3, and either $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$4 or $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$5, with $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$6 in weak coupling or $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$7 in strong coupling (Wang et al., 2020). The paper reports a threshold ratio $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$8 minimizing the nonreciprocal coupling required for perfect blockade under weak drive (Wang et al., 2020).

In the crossed-cavity gate system (Li et al., 30 Jul 2025), the strong-coupling example uses $$a_{j,out}(t) = a_{j,in}(t) - \sqrt{\kappa_j}\,\delta a_j(t)$$9s, $\omega_m$00s, and $\omega_m$01 MHz. The paper states that the protocols can operate in both weak and strong coupling regimes, but they rely on narrowband photons with

$\omega_m$02

so that the scattering amplitudes are effectively monochromatic (Li et al., 30 Jul 2025).

A common practical issue across these systems is that extra optical loss in the auxiliary or second cavity degrades the engineered effective channel. In the squeezing scheme, excess loss in cavity 2 increases $\omega_m$03 and reduces cooling (Wang et al., 2016). In the TLS-based gate scheme, cavity intrinsic loss and NV dephasing lower cooperativity and therefore reduce transmission contrast and gate performance (Li et al., 30 Jul 2025).

7. Conceptual significance and relation to adjacent cavity architectures

The double single-sided cavity system occupies a middle ground between strictly local single-cavity devices and fully symmetric double-sided resonators. Relative to a single cavity, it introduces an additional controlled mode or port without abandoning the one-channel input–output simplicity that makes scattering analysis tractable. Relative to a double-sided cavity, it avoids having two native external channels per resonator and instead allocates one port to each cavity, so inter-cavity transfer must be mediated internally.

This distinction matters in several neighboring literatures. In the KIN-gate analysis of double-sided cavities, naive injection from only one side fails deterministically because the atom couples only to a particular superposition of left- and right-incident modes, and the uncoupled fraction leads to a minimum failure probability (Gea-Banacloche et al., 2012). That complication is absent in architectures built directly from two single-sided cavities, where each port corresponds to a separate resonator and the internal mediator determines transmission or correlation (Li et al., 30 Jul 2025).

Likewise, the optomechanical works distinguish systems with direct optical tunneling from those with purely mechanical mediation. The photonic-molecule structure in (Wang et al., 2016) uses direct hopping $\omega_m$04 to engineer interference and effective linewidths, whereas the entanglement-transfer system in (Huan et al., 2015) has no direct tunneling and instead realizes a frequency-dependent, noise-bearing effective photon–photon interaction through $\omega_m$05. The literature therefore treats “double cavity” as a broader category, within which the “double single-sided” subclass is characterized not by a unique interaction term but by its port topology.

A plausible implication is that the enduring value of the double single-sided cavity system lies in modularity. One cavity can be optimized for strong intracavity interaction, another for narrowband filtering or low-loss extraction; one port can be used for drive, another for heralding or conditional scattering; and the mediator—mechanical mode, tunneling link, BEC collective excitation, or TLS—can be chosen according to the target functionality. Across squeezing (Wang et al., 2016), entanglement transfer (Huan et al., 2015), blockade (Wang et al., 2020), Fano engineering (Prakash et al., 2019), and photonic logic (Li et al., 30 Jul 2025), that modular separation of roles is the unifying operational principle.