Four-Mode Bosonic Josephson Junction

• Four-mode bosonic Josephson junctions are systems of four coupled bosonic modes that extend dimer physics into a network of interdependent phase dynamics.
• They encompass various experimental realizations—including photon condensates, quasi-periodic lattices, and composite junctions—each demonstrating unique behaviors such as synchronization and frustration.
• These systems enable controlled studies of many-body effects, offering insights into phase engineering, self-trapping, and dissipation-driven transitions.

A four-mode bosonic Josephson junction is a system of four coherent bosonic modes whose occupations and relative phases are coupled by Josephson tunneling. In the literature summarized here, that designation covers several non-equivalent but structurally related realizations: a rectangular network of four photon Bose–Einstein condensates connected by tunable $0$- and $\pi$-junctions, a four-localized-mode reduction of an atomic condensate in a quasi-periodic lattice, and a composite double-junction in continuous space whose many-body dynamics becomes four-fold fragmented. Across these settings, the defining ingredients are the coexistence of four mode amplitudes, phase-dependent particle exchange, and a multi-mode phase space supporting self-trapping, phase locking, frustration, and mode-coupled collective dynamics (Vretenar et al., 2020, Prates et al., 2022, Haldar et al., 2023).

1. Conceptual basis and canonical variables

The elementary building block of bosonic Josephson physics is the two-mode junction, in which the normalized population imbalance $z$ and relative phase $\phi$ form a canonical pair. In the internal bosonic Josephson junction realized with two coherently coupled hyperfine states of $^{87}\mathrm{Rb}$, the mean-field Hamiltonian is

$H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$

with $\Lambda=\chi N/\Omega$, where $\chi$ sets the nonlinear interaction scale and $\Omega$ the linear coupling scale. The same system admits a path-integral formulation in which $z$ and $\pi$0 are canonically conjugate, and the effective only-phase description is obtained by integrating out the imbalance variable (Zibold et al., 2010, Furutani et al., 2021).

A four-mode bosonic Josephson junction generalizes this structure by replacing a single imbalance–phase pair with several coupled collective coordinates. In the quasi-periodic lattice construction, the condensate wavefunction is projected onto four localized modes,

$\pi$1

and the natural dynamical variables become $\pi$2: imbalances and phase differences within the $\pi$3- and $\pi$4-pairs, an inter-pair population imbalance $\pi$5, and an inter-pair phase mismatch $\pi$6. The resulting phase space is therefore intrinsically higher-dimensional than that of a dimer, even before adding dissipation or external driving (Prates et al., 2022).

2. Hamiltonian structures in four-mode junctions

The four-mode Hamiltonian is not unique; it depends on whether the underlying platform is conservative, driven-dissipative, or explicitly coupled to auxiliary bosonic modes. In the quasi-periodic lattice realization, the four-mode reduction is Hamiltonian in rescaled time $\pi$7 and can be written as

$\pi$8

where $\pi$9 and $z$0 are modified two-mode Josephson Hamiltonians for the two localized pairs and $z$1 contains the inter-pair energy offset $z$2 and nonlinear cross-couplings proportional to $z$3, $z$4, and $z$5. In the decoupled limit $z$6, the model reduces to two independent Josephson junctions, each recovering the standard two-mode equations when all atoms occupy a single pair (Prates et al., 2022).

In the photon-condensate network, the appropriate description is driven-dissipative. For four condensates $z$7, the field amplitudes are $z$8, and a formal four-mode Hamiltonian can be written as

$z$9

with on-site energies $\phi$0 set by local cavity modes and potentials and complex Josephson couplings $\phi$1 determined by the optical landscape. Because the system is not conservative, the phase configuration is selected by gain maximization rather than by minimization of a bare static Hamiltonian (Vretenar et al., 2020).

A dissipative extension appears when a conventional two-mode junction is coupled to discrete bosonic bath modes. In that case, a four-mode junction arises naturally as two localized condensate modes plus two retained bath modes. In the bosonic language, the Hamiltonian takes the schematic form

$\phi$2

so that the Josephson mode hybridizes with additional bosonic degrees of freedom and the mode structure becomes genuinely multi-component (Sinha et al., 2019).

3. Dynamical regimes, fixed points, and bifurcations

The simplest bifurcation structure is already visible in the two-mode internal junction. For $\phi$3, the $\phi$4-phase fixed point $\phi$5 is stable. At $\phi$6, it undergoes a supercritical pitchfork bifurcation, and for $\phi$7 two new stable fixed points appear,

$\phi$8

signaling the transition from Rabi dynamics to Josephson dynamics with macroscopic self-trapping. The associated separatrix divides oscillatory and self-trapped trajectories in phase space (Zibold et al., 2010).

Four-mode systems inherit these two-mode regimes but add inter-mode energy exchange and slow collective modulations. In the quasi-periodic lattice realization, the weakly nonlinear regime supports coherent oscillations within the strongly coupled $\phi$9-junction together with a slow population exchange with the weakly coupled $^{87}\mathrm{Rb}$0-junction. This produces beating in $^{87}\mathrm{Rb}$1, a feature absent in a strict two-mode model. The same four-mode Hamiltonian also supports switching between self-trapped and oscillatory trajectories: as atoms move between the two mode pairs, the effective fixed-point conditions change, and trajectories can leave one self-trapped region of phase space and enter another or pass through wide oscillations (Prates et al., 2022).

A structurally related four-mode problem is the two-species double-well junction, whose dynamical variables are $^{87}\mathrm{Rb}$2. In that system, six scenarios of measure synchronization were identified: two in the Josephson-oscillation regime and four in the self-trapping regime. At the synchronization transition, the average energies of the two species merge, the phase-space domains explored by the two subsystems coincide, and the approach to the transition obeys a power law with critical exponent $^{87}\mathrm{Rb}$3. Poincaré sections show that the transition is associated with separatrix crossing, which the analysis presents as the general mechanism behind the different synchronization scenarios (Tian et al., 2013).

4. Representative realizations

The designation “four-mode bosonic Josephson junction” therefore covers several experimentally and theoretically distinct constructions.

Realization	Four modes	Characteristic feature
Photon Bose–Einstein condensates	Four corner condensates in a rectangular network	Tunable $^{87}\mathrm{Rb}$4- and $^{87}\mathrm{Rb}$5-junctions by thermo-optical control
Quasi-periodic lattice condensate	$^{87}\mathrm{Rb}$6	Two effective junctions coupled nonlinearly
Composite bosonic Josephson junction	Four wells in a continuous-space trap	Four-fold fragmentation and two-frequency tunneling
Two-species double well	$^{87}\mathrm{Rb}$7	Synchronization and separatrix-crossing dynamics

In the photon-condensate implementation, four condensates are created at the corners of a rectangular structure fabricated in a dye-filled microcavity. Neighboring condensates are linked by tunable Josephson junctions whose sign can be switched between $^{87}\mathrm{Rb}$8 and $^{87}\mathrm{Rb}$9 by patterned thermo-optical heating. Interferometric imaging confirms mutual coherence across the network, including between diagonally opposite condensates, and the resulting phase configurations are directly interpreted as four XY spins on a graph (Vretenar et al., 2020).

In the quasi-periodic lattice implementation, the four modes are not imposed by a geometric four-well trap but extracted from localized even and odd eigenstates below a mobility edge. The selected modes $H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$0 form two effective Josephson junctions with very different tunneling splittings, $H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$1. The cross-coupling parameters $H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$2, $H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$3, and $H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$4 arise from overlap integrals of the localized mode profiles and are responsible for the slow inter-pair population exchange that distinguishes the four-mode dynamics from a pair of uncoupled dimers (Prates et al., 2022).

The composite bosonic Josephson junction provides a continuum-space realization. There, each side of an ordinary double well is itself split by a Gaussian hump, so that the full trap supports four localized regions. Many-body simulations with MCTDHB show that $H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$5 orbitals are required for convergence, unlike the standard two-mode BJJ where $H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$6 is sufficient in the same weak-interaction regime. The system displays two-frequency tunneling and four-fold fragmentation, making it an explicit many-body realization of a four-mode junction rather than merely a few-mode truncation (Haldar et al., 2023).

5. Phase engineering, frustration, and effective spin models

A major development in four-mode junctions is the ability to engineer not only the magnitude but also the sign of the coupling. In the photon-condensate network, the effective total-gain equation for four condensates can be written as

$H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$7

with

$H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$8

If the photon numbers are equal, the second term is exactly the classical XY Hamiltonian

$H=\frac{\Lambda}{2}z^2-\sqrt{1-z^2}\cos\phi,$9

The four-condensate system then acts as a minimal multi-junction network whose phases “compute” the gain-maximizing, equivalently $\Lambda=\chi N/\Omega$0-minimizing, configuration. By choosing which links are $\Lambda=\chi N/\Omega$1-junctions and which are $\Lambda=\chi N/\Omega$2-junctions, one can realize frustrated phase patterns reminiscent of frustrated magnetic configurations (Vretenar et al., 2020).

Periodic driving offers a second route to phase engineering. In a two-mode bosonic Josephson junction under resonant Floquet modulation, the effective tunnel coupling is

$\Lambda=\chi N/\Omega$3

so both its magnitude and phase are controlled by the drive amplitude and phase. A plausible implication is that a four-mode array with independently modulated links could realize bond-dependent complex couplings, synthetic gauge fluxes on closed loops, and coupled sine-Gordon sectors with several relative phase fields, although that extension is presented as a foundation rather than as a completed four-mode calculation (Ji et al., 2022).

High-frequency modulation of the tunnel coupling yields a further possibility: ponderomotive stabilization of nontrivial phase states. In the driven two-mode junction, the effective potential generated by rapid modulation stabilizes a $\Lambda=\chi N/\Omega$4-phase mode through a Kapitza-pendulum mechanism, and, when the mean tunneling vanishes and the small-population-difference approximation fails, the momentum-shortening effect can stabilize a $\Lambda=\chi N/\Omega$5-phase mode. A plausible implication is that four-mode networks could combine $\Lambda=\chi N/\Omega$6-, $\Lambda=\chi N/\Omega$7-, and $\Lambda=\chi N/\Omega$8-links to realize more elaborate frustrated phase patterns, but the explicit four-mode problem is not worked out in detail (Lin et al., 2024).

6. Many-body correlations, dissipation, and significance

Four-mode bosonic Josephson junctions are not only classical phase models; they also provide a controlled setting for many-body fragmentation and dissipative mode hybridization. In the composite junction, the one-body density matrix develops four relevant natural orbitals, and the long-time occupation fractions $\Lambda=\chi N/\Omega$9 exhibit a universal degree of fragmentation for fixed interaction parameter $\chi$0: for finite hump height $\chi$1, the third and fourth orbitals become significantly occupied, and the fragmentation pattern is essentially independent of $\chi$2. Survival probability, position variance, and momentum variance all reveal physics that a fixed two-mode or Gross–Pitaevskii description misses (Haldar et al., 2023).

Dissipative four-mode variants arise when a Josephson junction is coupled to a finite bosonic environment. Retaining only two bath modes converts the problem into a four-mode system in which the junction mode hybridizes with the additional oscillators. In that setting, increasing the effective bath coupling drives a quantum phase transition at

$\chi$3

the lowest excitation gap closes, and number-imbalance fluctuations are enhanced. Dynamically, Josephson oscillations are damped, $\chi$4-oscillations and self-trapped states become unstable near resonances with bath modes, and with Ohmic dissipation they decay toward the ground state (Sinha et al., 2019).

The quantum-effective-action approach developed for the two-mode junction indicates how multi-mode quantum corrections may be organized. In the dimer, integrating out the imbalance yields an only-phase action, and the derivative expansion provides the quantum correction to the Josephson frequency. This suggests that a four-mode only-phase theory would involve several coupled phase variables and a matrix-valued kinetic term, although the explicit multi-phase construction is presented as a natural extension rather than as a fully derived result (Furutani et al., 2021).

Taken together, these results establish the four-mode bosonic Josephson junction as the minimal setting in which dimer physics gives way to genuinely networked behavior. It is large enough to support mode-coupled beating, switching, frustration, and four-fold fragmentation, yet still small enough to admit explicit Hamiltonian reductions and controlled experimental implementations. In photon condensates it already functions as a programmable $\chi$5 Josephson network and analog XY simulator, while in atomic systems it provides a bridge between conventional double-well Josephson physics and larger multi-junction arrays (Vretenar et al., 2020, Prates et al., 2022).