Hybrid-Liouvillian Formalism

• Hybrid-Liouvillian description is a framework that interpolates between conditional no-jump non-Hermitian evolution and full trace-preserving Lindblad dynamics.
• It uses a controllable parameter to blend coherent Hamiltonian motion with dissipative quantum jumps and monitor system trajectories.
• The formalism underpins analyses of exceptional points, transport phenomena, and integrability in both quantum open systems and related geometric models.

Hybrid-Liouvillian description denotes a family of formalisms in which Liouvillian evolution is not treated as a standalone GKSL generator, but is coupled to, or interpolated with, another dynamical sector. In recent open-quantum-system work, its central meaning is a continuous bridge between conditional no-jump evolution under an effective non-Hermitian Hamiltonian and unconditional Lindblad dynamics with quantum jumps (Minganti et al., 2020). Closely related usages combine coherent Hamiltonian motion with dissipative jump processes, embed Schrödinger–Liouville evolution and projective measurement into a single enlarged state space, or extend Liouville-type integrability to hybrid Hamiltonian systems with impacts (Krhac et al., 7 Apr 2025, López-Gordón et al., 2023). The common structural theme is the retention of Liouvillian or Liouville-type organization while admitting an additional component—jump recycling, monitoring, discrete resets, or geometric extension—that changes the relevant state space, spectrum, or invariant structure.

1. Core open-system meaning

In the open-system literature, the standard hybrid-Liouvillian construction introduces a continuous interpolation between the full trace-preserving master equation and the no-jump non-Hermitian limit. For a qubit with Hamiltonian $H$, jump operator $L$, and rate $\gamma$, one representative form is

$$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$

with $q\in[0,1]$ (Kumar et al., 2021). An equivalent convention used in monitored-qubit work is

$$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$

again with $q\in[0,1]$ (Paul et al., 13 May 2026).

These forms encode the same limiting structure. At $q=1$, the dynamics reduces to standard trace-preserving GKSL/Lindblad evolution. At $q=0$, the recycling term is removed and the evolution becomes the no-jump, post-selected, non-Hermitian limit, generated by an effective Hamiltonian such as

$$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$

depending on convention (Kumar et al., 2021, Paul et al., 13 May 2026). For intermediate $L$0, only part of the jump contribution is retained.

In the more general superoperator language, the hybrid generator is written as an interpolation between the full Liouvillian $L$1 and the jump-free no-jump Liouvillian $L$2,

$L$3

so the parameter $L$4 directly controls how much of the quantum-jump sector remains in the effective description (Minganti et al., 2020). This construction is the principal contemporary meaning of “hybrid-Liouvillian formalism.”

2. Quantum trajectories, postselection, and detector efficiency

The hybrid parameter acquires its physical meaning from quantum trajectories. In that picture, the Lindblad equation is the ensemble average over stochastic trajectories with random jump events, whereas the no-jump effective Hamiltonian describes trajectories conditioned on the absence of detected jumps (Minganti et al., 2020). The hybrid-Liouvillian description interpolates between these two ensemble constructions.

Several papers make this operational interpretation explicit. One formulation states that whenever a photon is emitted, the run is discarded with probability $L$5, or allowed to continue with probability $L$6 (Kumar et al., 2021). Another identifies $L$7 as the detector-efficiency or jump-retention parameter, i.e. the fraction of quantum-jump trajectories retained in the conditioned ensemble (Paul et al., 13 May 2026). In the finite-efficiency detector picture of the original hybrid-Liouvillian formalism, one has $L$8, with $L$9 the detector efficiency; the same paper also gives a two-detector construction in which a beam splitter sends a fraction $\gamma$0 of jumps to one detector and $\gamma$1 to the other, and postselection on one detector realizes the hybrid generator (Minganti et al., 2020).

For $\gamma$2, the state is generally not trace preserving. The monitored-qubit analysis therefore evaluates observables on the normalized state

$\gamma$3

and emphasizes that normalization feeds trace decay back into the Bloch-vector dynamics, making the effective evolution nonlinear (Paul et al., 13 May 2026). This nonlinear renormalization is not a peripheral detail; it controls measurable temporal correlations. In the Leggett–Garg setting, the three-time quantity

$\gamma$4

approaches its algebraic maximum $\gamma$5 as $\gamma$6, while any finite increase in $\gamma$7 produces a rapid, highly nonlinear suppression; the paper fits the maximal value by

$\gamma$8

(Paul et al., 13 May 2026).

The same interpolation is used as a control resource in exceptional-point protocols. For a decaying qubit, adiabatic encircling in the hybrid-Liouvillian regime can combine pure-state conversion with reduced loss, while a hopping protocol can ideally achieve

$\gamma$9

with $$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$0 the conversion fidelity and $$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$1 the postselection success probability (Kumar et al., 2021). In this sense, detector efficiency, postselection, and trajectory conditioning are not merely interpretive layers; they are the operational backbone of the hybrid-Liouvillian description.

3. Exceptional points in Hamiltonian, Liouvillian, and hybrid spectra

A major application of hybrid-Liouvillian descriptions is the comparison between Hamiltonian exceptional points (HEPs) of non-Hermitian effective Hamiltonians and Liouvillian exceptional points (LEPs) of superoperators acting on density matrices. The brief review on LEP encircling defines the Liouvillian by

$$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$2

and emphasizes that LEPs can modify both steady-state structure and transient dynamics (Sun et al., 2024).

The hybrid-Liouvillian formalism shows that HEPs and LEPs need not coincide. In one spin-$$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$3 model,

$$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$4

the effective Hamiltonian is diagonal and has no HEP, whereas the full Liouvillian has an EP at $$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$5. The hybrid spectrum moves this EP to

$$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$6

so the EP disappears in the no-jump limit $$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$7 and is fully realized only when quantum jumps are included (Minganti et al., 2020). In a second qubit model,

$$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$8

the effective Hamiltonian has an HEP at $$\frac{d\rho}{dt}=\mathcal{L}_{q}[\rho] =-i[H,\rho]-\frac{\gamma}{2}\left(\{L^\dagger L,\rho\}-2q\,L\rho L^\dagger\right),$$9, while the full Liouvillian has an LEP at $q\in[0,1]$0; moreover, the $q\in[0,1]$1 hybrid Liouvillian exhibits a third-order EP in Liouville space (Minganti et al., 2020).

Optomechanics provides a sharper separation between conditional and unconditional singularities. For the red-sideband cavity-optomechanical model, the unconditional drift matrix

$q\in[0,1]$2

has a Liouvillian exceptional point at

$q\in[0,1]$3

which is temperature independent. By contrast, the no-jump non-Hermitian matrix

$q\in[0,1]$4

has a Hamiltonian exceptional point at

$q\in[0,1]$5

which is temperature dependent (Ghosh et al., 25 Feb 2026). The same work introduces a jump-weight parameter $q\in[0,1]$6,

$q\in[0,1]$7

and derives a continuous family of hybrid exceptional points $q\in[0,1]$8 satisfying

$q\in[0,1]$9

The second-order onset near $$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$0 is the paper’s robustness result for weak jump contamination (Ghosh et al., 25 Feb 2026).

Atomic-vapor calculations reach a similar conclusion from the opposite direction. There the hybrid-Liouvillian interpolation is used to show that quantum jumps can shift, lift, create, or destroy spectral degeneracies, and can reduce the order of an EP relative to the jump-free prediction (Kopciuch et al., 3 Jun 2025). Analytical classification can then be pushed further with Newton polygons and tropical geometry: for Liouvillian characteristic polynomials, slopes of Newton-polygon segments determine Puiseux exponents, so the same degeneracy can split with $$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$1, $$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$2, or linear scaling depending on the perturbation direction (P et al., 9 Oct 2025). The hybrid-Liouvillian description is therefore a spectral bridge, not only a dynamical interpolation.

4. Coherent–dissipative transport, topology, and Liouvillian skin phenomena

A second major usage of hybrid Hamiltonian–Liouvillian description concerns systems whose coherent Hamiltonian sector remains explicit, while nonreciprocity, disorder, or decoherence is encoded at the Liouvillian level. In the one-dimensional lattice model of erratic Liouvillian skin localization, the coherent part is

$$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$3

while incoherent hopping is generated by

$$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$4

with

$$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$5

For Bernoulli-distributed $$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$6, the model is globally reciprocal but locally asymmetric: conventional boundary accumulation disappears, the steady state becomes erratically localized in the bulk, and in the strong-dissipation regime $$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$7 the dynamics reduces to a Sinai random walk with

$$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$8

(Longhi, 16 Feb 2026). The result is a direct statement that “no skin effect” in the boundary-accumulation sense does not imply normal transport.

Hybrid Hamiltonian–Liouvillian interplay also appears in relaxation spectroscopy. In the Kitaev chain with local dephasing, weak dissipation yields a Liouvillian gap

$$\frac{d\rho}{dt} = -i[H,\rho] + 2\gamma\left( q\,L\rho L^\dagger -\frac12\{L^\dagger L,\rho\} \right),$$9

which becomes independent of $q\in[0,1]$0 in the topological phase $q\in[0,1]$1, while in the non-topological phase $q\in[0,1]$2 it is suppressed for large $q\in[0,1]$3. In the strong-dissipation regime, by contrast, the gap is approximately

$q\in[0,1]$4

and the topology of the underlying Hamiltonian becomes largely irrelevant at leading order (Kavanagh et al., 2024). The Liouvillian gap here is a hybrid diagnostic: it reflects both dissipation and band topology, but only in the regime where coherent structure survives.

Photonic quantum walks provide a channel-based realization of the same idea. The coherent step operator

$q\in[0,1]$5

is combined with stochastic phase noise, and the ensemble-averaged density matrix evolves as

$q\in[0,1]$6

with coherence parameter

$q\in[0,1]$7

The model interpolates between the fully coherent quantum walk $q\in[0,1]$8 and the fully incoherent classical limit $q\in[0,1]$9. The measured center-of-mass drift exhibits a crossover around $q=1$0: for weak non-Hermiticity, coherence enhances transport, whereas for stronger non-Hermiticity decoherence enhances transport; the same platform shows interface accumulation and a long-time drift determined by the instantaneous channel (Yang et al., 25 Jun 2026).

In two-dimensional electron systems, the hybrid structure is built from a coherent Rashba Hamiltonian and finite-temperature GKSL band-transition processes. There the Liouvillian supports $q=1$1 and $q=1$2 skin effects, strongest below the band-splitting scale $q=1$3, together with a scale-free localization length

$q=1$4

so that the relaxation time saturates to $q=1$5 in the thermodynamic limit (Shigedomi et al., 23 May 2025). A closely related synthetic-state-space viewpoint reinterprets optical pumping as Liouvillian skin effect: the target state acts as an open boundary, OBC and PBC in state space yield sharply different spectra, and the Liouvillian gap controls the pumping and cooling rate (Cai et al., 2024).

5. Measurement, projective dynamics, and refined hybrid state spaces

A distinct but structurally related line of work uses “hybrid Schrödinger–Liouville” to eliminate the usual piecewise split between smooth evolution and instantaneous collapse. The refined state space is

$q=1$6

where $q=1$7 is a $q=1$8-dimensional Hilbert space and $q=1$9 is a discrete classical register (Krhac et al., 7 Apr 2025). From $q=0$0, one defines the induced quantum state

$q=0$1

the classical probabilities

$q=0$2

and the conditional collapsed states

$q=0$3

The most general completely positive linear dynamics preserving normalization are written as the Poulin equation,

$q=0$4

with positivity conditions $q=0$5 and $q=0$6 on the coefficients $q=0$7 (Krhac et al., 7 Apr 2025). Schrödinger–Liouville evolution is recovered by a special choice of coefficients, yielding

$q=0$8

Projective measurement is embedded into the same continuous framework. For the simplest exactly solvable choice,

$q=0$9

the hybrid measurement dynamics become

$$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$0

with solution

$$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$1

Consequently,

$$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$2

and

$$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$3

The formulation therefore reproduces outcome probabilities and the averaged post-measurement state exponentially in time, while treating measurement as part of one differential equation rather than as an external jump rule (Krhac et al., 7 Apr 2025).

The same paper explicitly restricts this framework to the “ex ante” measurement postulate: it captures the probabilities of outcomes and the corresponding averaged post-measurement quantum state, but not the “ex post” state update conditioned on the actually observed outcome. Its stated motivation is control-theoretic. By encoding quantum and classical information in one object $$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$4, the construction is made amenable to standard control-theoretic analysis, port-Hamiltonian formulations, and Dirac-structure-based interconnections (Krhac et al., 7 Apr 2025).

6. Geometric and integrability extensions

Outside GKSL interpolation, Liouvillian language also appears in hybrid Hamiltonian systems, differential-algebraic extensions, propagator theory, and symplectic integration. In the hybrid Hamiltonian setting, a hybrid system is the 4-tuple

$$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$5

with smooth manifold $$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$6, vector field $$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$7, impact surface $$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$8, and impact map $$\tilde H=H-i\frac{\gamma}{2}L^\dagger L \quad\text{or}\quad H_{\rm eff}=H-i\gamma L^\dagger L,$$9. The dynamics are

$L$00

with the convention that after impact the state is $L$01 (López-Gordón et al., 2023). For hybrid Hamiltonian systems $L$02, the paper introduces generalized hybrid constants of motion: functions $L$03 satisfying $L$04 and whose post-impact value depends only on the pre-impact value on each connected component of the switching surface. A completely integrable hybrid Hamiltonian system $L$05 then obeys the hybrid analogue of Liouville integrability. Regular common level sets

$L$06

are Lagrangian; compact connected components are diffeomorphic to $L$07; and impacts map one invariant leaf to another,

$L$08

In action-angle coordinates,

$L$09

while the action variables after impact depend only on the pre-impact actions (López-Gordón et al., 2023).

In geometric control, Liouvillian systems are defined as extensions of flat systems by quadratures and exponentials of integrals. Using diffieties and infinite prolongation, a differential extension $L$10 of a flat system $L$11 is Liouvillian if there exists a chain

$L$12

such that each step is either an integral extension

$L$13

or an exponential-of-integral extension

$L$14

with $L$15 (Chelouah, 2010). Here the “hybrid” structure is the decomposition into a flat subsystem and a complementary Liouvillian layer.

Liouvillian ideas also organize exact propagation. For time-dependent quadratic Schrödinger equations, the propagator is reconstructed from a Riccati system and its associated characteristic equation,

$L$16

The paper’s main theorem states that

$L$17

so Liouvillian solvability of the characteristic equation is equivalent to explicit Liouvillian propagators (Acosta-Humánez et al., 2013). At a more algebraic level, the spectral set of second-order linear ODEs with polynomial coefficients that admit Liouvillian solutions is described as a countable union of disjoint algebraic varieties $L$18, and for polynomial Schrödinger potentials the number of Liouvillian eigenvalues is bounded by $L$19 (Acosta-Humánez et al., 2019).

Finally, Liouvillian forms on exact symplectic manifolds are used to construct midpoint-type implicit symplectic integrators. On $L$20, a Liouville vector field $L$21 satisfies $L$22, with associated Liouvillian form $L$23. The resulting numerical scheme has the form

$L$24

and is presented as a symmetric modification of the symplectic midpoint rule obtained from a Liouvillian-form-induced isotopy of the exact Hamiltonian flow toward the segment joining two consecutive points (Jiménez-Pérez, 2015).

These mathematical extensions do not identify Liouvillian dynamics with GKSL superoperators. Rather, they retain the broader Liouville/Liouvillian emphasis on invariant geometric structure, quadrature-based reconstruction, or canonical 1-form geometry. This suggests that “hybrid-Liouvillian description” has developed into a wider organizing label: in quantum open systems it interpolates between no-jump and full-jump dynamics, while in adjacent mathematical settings it denotes the preservation of Liouville-type structure under added resets, measurements, impacts, or differential extensions.