Driven Tavis-Cummings Model Overview

• Driven Tavis–Cummings model is a framework where external driving fields alter the symmetry, excitation structure, and control tasks of collective light–matter systems.
• It encompasses various methodologies—coherent driving, parametric pumping, frequency modulation, and Landau-Zener sweeps—that yield distinct spectral and dynamical responses.
• The model enables engineered interactions for quantum control and metrology by harnessing collective enhancements and driven superradiant phase transitions.

The driven Tavis–Cummings model is the class of Tavis–Cummings systems in which the ensemble–cavity dynamics is modified by an external drive or explicit time dependence. In the cited literature, this includes weak coherent spectroscopy of a cavity mode, off-resonant cavity driving in a rotating frame, two-photon parametric driving, modulation of the spin frequency, and linear sweeps of the bosonic mode frequency. Across these formulations, the central object remains an ensemble of two-level systems or an effective collective spin coupled to a single bosonic mode, but the drive changes the symmetry, excitation-number structure, spectral response, and accessible control tasks (0812.2651, Feng et al., 2014, Lü et al., 2023, Sun et al., 2016).

1. Hamiltonian structure and driving conventions

The standard Tavis–Cummings Hamiltonian in the rotating-wave approximation is

$$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \sum_{j=1}^{N} \frac{\omega_{q,j}}{2}\, \sigma_j^z + \sum_{j=1}^{N} g_j \left(a\,\sigma_j^+ + a^\dagger\,\sigma_j^- \right),$$

with cavity operators $a,a^\dagger$, qubit frequencies $\omega_{q,j}$, Pauli operators $\sigma_j^{z,\pm}$, and couplings $g_j$ (0812.2651). For identical qubits and uniform couplings, one introduces

$$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$

so that

$$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$

For non-identical but fixed couplings, a useful effective collective coupling is

$G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$

which reduces to $g\sqrt{N}$ for $g_j=g$ (0812.2651).

A coherent cavity drive is commonly added as

$a,a^\dagger$0

or, in an equivalent convention,

$a,a^\dagger$1

with drive frequency $a,a^\dagger$2 and amplitude $a,a^\dagger$3 (0812.2651, Blaha et al., 2021). In a frame rotating at $a,a^\dagger$4, one obtains a driven Hamiltonian of the form

$a,a^\dagger$5

with $a,a^\dagger$6 and $a,a^\dagger$7 (0812.2651). In the circuit-QED implementation of a driven Tavis–Cummings transition, the homogeneous rotating-frame Hamiltonian is written as

$a,a^\dagger$8

where the off-resonant drive renormalizes the relevant energy scale from bare frequencies to detunings (Feng et al., 2014).

A second major driven variant is the parametrically driven model, where the cavity acquires a two-photon term. In one rotating-frame formulation,

$a,a^\dagger$9

with cavity loss described by

$\omega_{q,j}$0

(Geng et al., 31 Aug 2025). A related static effective form is

$\omega_{q,j}$1

which retains the rotating-wave spin–photon coupling but adds a static squeezing term (Lü et al., 2023).

Other driven formulations are explicitly time dependent. One is the beyond-RWA model

$\omega_{q,j}$2

where the drive is implemented by modulating the spin frequency, not by an explicit cavity-drive term (Ovsiannikov et al., 23 Apr 2026). Another is the Landau–Zener driven Tavis–Cummings model,

$\omega_{q,j}$3

in which the boson frequency is swept linearly in time (Sun et al., 2016).

This diversity of Hamiltonians shows that “driven Tavis–Cummings model” is not a single unique operator. The common feature is the Tavis–Cummings light–matter sector; the distinction lies in whether the drive acts as a coherent source, a parametric pump, a control-field modulation, or a time-dependent frequency sweep.

2. Collective dressed states, bright and dark manifolds, and spectroscopy

In the single-excitation manifold of the resonant Tavis–Cummings model, the cavity couples only to a collective bright state. For identical couplings,

$\omega_{q,j}$4

while the remaining $\omega_{q,j}$5 orthogonal combinations are dark states $\omega_{q,j}$6 satisfying $\omega_{q,j}$7 (0812.2651). In the basis spanned by the one-photon state and the bright qubit excitation, the resonant single-excitation Hamiltonian is

$\omega_{q,j}$8

so the dressed polariton states are

$\omega_{q,j}$9

with energies

$\sigma_j^{z,\pm}$0

The vacuum Rabi splitting is therefore

$\sigma_j^{z,\pm}$1

which is the discrete $\sigma_j^{z,\pm}$2 enhancement measured in circuit QED for one, two, and three transmon qubits strongly coupled to a resonator (0812.2651).

Within the single-excitation manifold, the bright state is the $\sigma_j^{z,\pm}$3-state,

$\sigma_j^{z,\pm}$4

so cavity spectroscopy probes the bright $\sigma_j^{z,\pm}$5-state component but not the orthogonal dark combinations (0812.2651). In the actual device of Fink et al., the standing-wave electric field changes sign between the resonator center and its ends, giving $\sigma_j^{z,\pm}$6; the bright state is then weighted by the coupling signs, while the effective collective coupling remains $\sigma_j^{z,\pm}$7 (0812.2651).

Under weak driving, the driven transmission spectrum shows the two bright polariton peaks

$\sigma_j^{z,\pm}$8

and an avoided crossing as the qubits are tuned through the cavity (0812.2651). Dark states do not appear in transmission at degeneracy because they have zero matrix element with the cavity operator under uniform couplings. For two qubits, the antisymmetric state

$\sigma_j^{z,\pm}$9

is absent in transmission; for three qubits, the two dark states $g_j$0 are likewise not visible at degeneracy (0812.2651).

A system-theoretic treatment recasts the single-excitation driven Tavis–Cummings problem as a linear passive system. For equal transition frequencies, the one-port transfer function is

$g_j$1

and on resonance

$g_j$2

with symmetric peaks at $g_j$3 (Dong et al., 2021). In that formulation, the bright two-mode subsystem is controllable and observable, whereas the dark manifold is uncontrollable and unobservable and forms a decoherence-free subsystem (Dong et al., 2021).

A recurrent misconception is that dark states disappear from the model itself. The cited works show a narrower statement: dark states remain present in the spectrum, but a uniform cavity drive cannot excite them at exact symmetry, so they are absent from transmission spectroscopy rather than absent from the Hilbert space (0812.2651, Dong et al., 2021).

3. Symmetry breaking, nonequilibrium criticality, and superradiant regimes

For the undriven Tavis–Cummings model, the total excitation number is conserved and the model has a continuous $g_j$4 symmetry (Feng et al., 2014, Lü et al., 2023). This changes qualitatively under driving, but the mechanism depends on the drive.

Under an off-resonant coherent cavity drive, the rotating-frame Hamiltonian

$g_j$5

breaks excitation-number conservation and parity (Feng et al., 2014). In a four-qubit circuit, this permitted a non-equilibrium quantum phase transition to be traversed dynamically by sweeping $g_j$6 while holding $g_j$7 fixed. Linearization around the normal phase yields the renormalized critical condition

$g_j$8

or $g_j$9 when both detunings are negative, as in the experiment (Feng et al., 2014). The principal observed signature was an abrupt rise of the scaled moment $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$0 once $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$1 crossed unity, in agreement with Lindblad simulations including $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$2, $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$3, and $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$4 (Feng et al., 2014).

Parametric driving changes the symmetry in a different way. Adding the two-photon term

$$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$5

or equivalently

$$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$6

breaks the Tavis–Cummings $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$7 symmetry down to a discrete $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$8 symmetry (Geng et al., 31 Aug 2025, Lü et al., 2023). In this setting, superradiance is associated with $$S^\pm = \sum_{j=1}^N \sigma_j^\pm, \qquad S_z = \frac{1}{2}\sum_{j=1}^N \sigma_j^z,$$9 symmetry breaking, finite cavity coherence $$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$0, finite transverse spin polarization, and nonzero anomalous correlator $$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$1 due to squeezing (Geng et al., 31 Aug 2025). For the symmetric parametrically driven model,

$$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$2

the superradiant phase transition occurs at

$$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$3

with a second-order ground-state transition, closed-form low-energy eigenstates, and a quantum metric that diverges as the excitation gap closes (Lü et al., 2023).

When all-to-all exchange interactions are added,

$$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$4

the parametric superradiant threshold is modified to

$$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$5

in the thermodynamic Holstein–Primakoff limit (Geng et al., 31 Aug 2025). Repulsive $$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$6 suppresses superradiance monotonically, whereas attractive $$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$7 can lower the threshold but also create “finger-like” normal regions at finite $$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$8 when the lowest Dicke-level gap is locally maximized (Geng et al., 31 Aug 2025).

These driven formulations demonstrate that the phrase “superradiant transition in a Tavis–Cummings model” must be interpreted with care. In the cited works, the transition is enabled either by coherent driving that breaks excitation-number conservation in the rotating frame or by a parametric two-photon term that reduces $$\frac{H_{\mathrm{TC}}}{\hbar} = \omega_c\, a^\dagger a + \omega_q\, S_z + g\,\left(a\,S^+ + a^\dagger\,S^-\right).$$9 to $G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$0; it is not a property of the equilibrium rotating-wave Tavis–Cummings Hamiltonian by itself (Feng et al., 2014, Lü et al., 2023, Geng et al., 31 Aug 2025).

4. Time-dependent, exactly solvable, and structure-preserving driven regimes

A distinct driven branch is the Landau–Zener extension, where the boson frequency varies linearly in time. The driven Hamiltonian

$G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$1

conserves the total number of excitations

$G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$2

so the evolution decomposes into fixed-$G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$3 blocks (Sun et al., 2016). This model is an exactly solvable multistate Landau–Zener problem. At each ordered crossing, the stay probability is

$G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$4

and the full transition probability factorizes into a product of such two-level factors (Sun et al., 2016). For fully polarized initial states, the coarse-grained transition probabilities are expressed in terms of $G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$5-Pochhammer symbols and $G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$6-binomial coefficients, revealing a direct connection between the driven Tavis–Cummings model and $G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$7-deformed binomial statistics (Sun et al., 2016).

Another time-dependent branch abandons excitation-number conservation altogether by retaining counter-rotating terms and modulating the spin frequency:

$G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$8

Here the Hamiltonian is split as

$G_N = \sqrt{\sum_{j=1}^{N} g_j^2},$9

with

$g\sqrt{N}$0

A structural result is that $g\sqrt{N}$1 becomes tridiagonal when the basis is ordered by the conserved quantity $g\sqrt{N}$2, while $g\sqrt{N}$3 becomes tridiagonal when ordered by $g\sqrt{N}$4; switching between the two orderings is a permutation of basis elements (Ovsiannikov et al., 23 Apr 2026).

This permits a symplectic split-operator propagation of the time-ordered exponential:

$g\sqrt{N}$5

which is second-order accurate (Ovsiannikov et al., 23 Apr 2026). In its Cayley realization,

$g\sqrt{N}$6

each tridiagonal step costs $g\sqrt{N}$7 time and $g\sqrt{N}$8 memory, where $g\sqrt{N}$9 is the truncated Hilbert-space dimension (Ovsiannikov et al., 23 Apr 2026). The method was benchmarked against direct QuTiP solutions and reported asymptotic scalings of $g_j=g$0 for the Cayley/Thomas scheme and approximately $g_j=g$1 for block-diagonal exponentiation (Ovsiannikov et al., 23 Apr 2026).

These two driven sectors illustrate different meanings of solvability. In the Landau–Zener problem, solvability refers to exact scattering probabilities in closed form; in the modulated beyond-RWA problem, it refers to a basis structure that preserves unitarity and linear complexity under numerical time propagation (Sun et al., 2016, Ovsiannikov et al., 23 Apr 2026).

5. Open-system response, input–output theory, and the limits of the bright-mode picture

Driven Tavis–Cummings systems are frequently treated as open quantum systems. In the linear weak-excitation regime, a driven cavity–ensemble obeys

$g_j=g$2

$g_j=g$3

and with the bright collective mode

$g_j=g$4

the steady-state cavity field is

$g_j=g$5

(Blaha et al., 2021). This is the standard driven Tavis–Cummings response.

The cited analysis of optically dense ensembles shows that this reduction is not universally valid. The Tavis–Cummings description requires

$g_j=g$6

where $g_j=g$7 is the cavity free spectral range and $g_j=g$8 is the scattering rate into non-cavity modes (Blaha et al., 2021). In terms of waveguide-QED branching ratio $g_j=g$9 and atom number $a,a^\dagger$00, the dominant validity criterion becomes $a,a^\dagger$01 for a chiral ring and $a,a^\dagger$02 for a Fabry–Perot, equivalent to low single-pass optical depth (Blaha et al., 2021). When this fails, photons undergo successive scattering and the single instantaneous bright-mode description breaks down.

The generalized high-optical-depth description is a cascaded model in which the ensemble enters through the single-pass transmission

$a,a^\dagger$03

and, for a single-ended ring, the reflection spectrum is

$a,a^\dagger$04

In this regime, the atom-induced roundtrip phase can exceed $a,a^\dagger$05 multiple times, creating additional resonances near the atomic line and preventing the saturation of mode shifts at $a,a^\dagger$06 predicted by multimode Tavis–Cummings models (Blaha et al., 2021).

Bi-directional coherent driving produces another open-system phenomenon: coherent perfect absorption in a two-emitter Tavis–Cummings cavity with dipole–dipole interaction $a,a^\dagger$07 (Wang et al., 2021). For identical emitters in a symmetric cavity, the intracavity field is

$a,a^\dagger$08

with input–output relations

$a,a^\dagger$09

(Wang et al., 2021). Coherent perfect absorption requires phase-locked amplitude-matched inputs and the constraints

$a,a^\dagger$10

Because the DDI shifts the bright-state detuning to $a,a^\dagger$11, strong coupling and strong DDI can generate two tunable CPA frequencies not attainable with a single emitter under comparable detuning (Wang et al., 2021).

The open-system literature therefore qualifies the simplest bright/dark picture in two ways. First, dark-state protection and $a,a^\dagger$12 collective response remain accurate in low-excitation, low-optical-depth, symmetry-preserving settings. Second, strong propagation effects, mirror asymmetry, two-sided driving, or direct emitter–emitter couplings can qualitatively alter the driven response without abandoning the broader Tavis–Cummings framework (Blaha et al., 2021, Wang et al., 2021).

6. Control tasks, metrology, and engineered effective interactions

Driven Tavis–Cummings systems are used not only for spectroscopy but also for quantum control. A central example is the engineering of one-axis twisting from a dispersive driven Tavis–Cummings model. Starting from

$a,a^\dagger$13

and choosing $a,a^\dagger$14, a Schrieffer–Wolff transformation in the dispersive regime yields

$a,a^\dagger$15

The undesired Stark term $a,a^\dagger$16 entangles spin and boson sectors and degrades squeezing and GHZ-state generation, but the cited work shows that it can be dynamically cancelled by drive design (Liu et al., 12 Mar 2026).

For a constant drive, fast rotation about $a,a^\dagger$17 averages the dynamics to

$a,a^\dagger$18

provided $a,a^\dagger$19 (Liu et al., 12 Mar 2026). For a time-varying square-wave drive with small duty cycle and pulse-area condition $a,a^\dagger$20, the effective Hamiltonian becomes

$a,a^\dagger$21

with the full twisting rate $a,a^\dagger$22 and improved robustness to decoherence (Liu et al., 12 Mar 2026). In the corresponding spin-only master equations derived after cavity elimination, the time-varying scheme exhibits more favorable dissipative structure for squeezing than the constant-drive scheme (Liu et al., 12 Mar 2026).

Metrological applications exploit a different dispersive effective Hamiltonian,

$a,a^\dagger$23

where the unknown weak field $a,a^\dagger$24 enters $a,a^\dagger$25 and $a,a^\dagger$26 (Su et al., 2024). The corresponding generator of $a,a^\dagger$27-encoding is

$a,a^\dagger$28

With coherent light and a spin-coherent atomic state, the quantum Fisher information scales as

$a,a^\dagger$29

in the regime $a,a^\dagger$30, giving Heisenberg scaling in average photon number (Su et al., 2024). If either photon-number fluctuations or spin-number fluctuations are enhanced by squeezing, the cited work proves double-Heisenberg scaling,

$a,a^\dagger$31

for squeezed-vacuum optical input or one-axis-twisted atomic input (Su et al., 2024).

Critical metrology provides another route. In the parametrically driven model with

$a,a^\dagger$32

the quantum metric

$a,a^\dagger$33

diverges at the superradiant critical point because the gap closes, and a homodyne-based protocol exploits this critical enhancement without adiabatic preparation (Lü et al., 2023).

The control perspective returns to the earliest circuit-QED realization as well. There, cavity spectroscopy of one, two, and three qubits demonstrated globally and locally controllable access to collective states, including the bright $a,a^\dagger$34-state sector, in a system with fixed atom number and fixed positions (0812.2651). This suggests why driven Tavis–Cummings architectures recur across quantum information, quantum optics, and metrology: the same collective light–matter coupling can be used to reveal dressed-state structure, engineer effective nonlinearities, and amplify parameter sensitivity, provided the symmetry class and dissipation mechanism of the chosen drive are correctly identified (0812.2651, Liu et al., 12 Mar 2026, Su et al., 2024).