Time-Local Hybrid Liouvillian Dynamics

• Time-local hybrid Liouvillians are open-system generators that combine non-Hermitian no-jump evolution with a weighted recycling superoperator, bridging conditional quantum-trajectory and unconditional Lindblad dynamics.
• They enable controlled interpolation between trace-decaying dynamics and fully trace-preserving evolution, supporting advanced quantum measurement and state conversion protocols.
• The framework reveals subtle spectral features including exceptional points and non-Markovian effects, highlighting the critical role of continuous measurement and state renormalization.

Time-local hybrid Liouvillians are open-system generators that combine the effective non-Hermitian no-jump evolution with a weighted recycling or jump superoperator, thereby interpolating between conditional quantum-trajectory dynamics and unconditional Lindblad dynamics. In representative form,

$$\frac{d\rho}{dt} = -\,i\bigl(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger\bigr) + \chi\,\sum_k L_k\rho L_k^\dagger, \qquad H_{\rm eff}=H-\frac{i}{2}\sum_k L_k^\dagger L_k,$$

where the interpolation parameter $\chi$ is written as $q$, $\epsilon$, or $\eta$ depending on the convention. In the qubit model used for Leggett–Garg analysis, the same structure appears as

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

with $q\in[0,1]$ interpreted as detector efficiency or post-selection weight (Paul et al., 13 May 2026). Across the recent literature, the construction serves as a time-local bridge between non-Hermitian Hamiltonian descriptions, quantum-jump unravelings, Liouvillian spectra, exceptional points, and continuously monitored dynamics (Minganti et al., 2020).

1. Formal decomposition of the generator

The formal starting point is the standard Lindblad master equation

$$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$

Introducing

$$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$

splits the dynamics into a non-Hermitian smooth part and a recycling superoperator,

$$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$

The hybrid construction then takes the form

$\chi$0

or, in optomechanics,

$\chi$1

(Kopciuch et al., 3 Jun 2025, Ghosh et al., 25 Feb 2026).

In the two-level Leggett–Garg setting one chooses

$\chi$2

so that

$\chi$3

and the full generator can be written as

$\chi$4

(Paul et al., 13 May 2026).

A time-dependent version appears in hybrid-Liouvillian state-conversion protocols, where

$\chi$5

with $\chi$6 used as a controlled interpolation between trace-decaying non-Hermitian evolution and trace-preserving Lindbladian evolution (Kumar et al., 2021).

2. Interpolation parameters and limiting regimes

The central idea is uniform, but the interpolation parameter is not standardized. Some papers weight the jump term directly, while others use a detector-efficiency variable with the opposite limiting assignment.

Reference	Parameter	Limiting cases
(Paul et al., 13 May 2026)	$\chi$7	$\chi$8 no-jump; $\chi$9 Lindblad
(Ghosh et al., 25 Feb 2026)	$q$0	$q$1 no-jump; $q$2 Lindblad
(Kopciuch et al., 3 Jun 2025)	$q$3	$q$4 NHH; $q$5 full Liouvillian
(Kumar et al., 2021)	$q$6	$q$7 nHH; $q$8 Lindbladian
(Minganti et al., 2020)	$q$9	$\epsilon$0 no-jump; $\epsilon$1 full Lindblad

In the Leggett–Garg qubit model, $\epsilon$2 is the fraction of quantum-jump trajectories that are kept in the ensemble. The limits are operationally explicit: $\epsilon$3 corresponds to unobserved environment and standard trace-preserving Lindblad dynamics, while $\epsilon$4 corresponds to perfect null-measurement post-selection, where trajectories containing a jump are discarded and only no-jump runs are retained (Paul et al., 13 May 2026).

In the 2020 post-selection formulation, $\epsilon$5 is the detector efficiency, defined as the fraction of jumps that are unambiguously recorded. Post-selecting on runs with no recorded jump yields

$\epsilon$6

Thus $\epsilon$7 gives purely no-jump evolution, whereas $\epsilon$8 reproduces the full Lindblad master equation (Minganti et al., 2020).

These conventions encode the same structural dichotomy: one endpoint is conditional non-Hermitian dynamics without recycling, and the other is the unconditional trace-preserving Liouvillian with full jump back-action.

3. Time-locality, divisibility, and normalization

The defining feature of the framework is time locality. In the qubit construction of (Paul et al., 13 May 2026), the generator is time-independent, so the map

$\epsilon$9

is time-local and infinitesimally divisible:

$\eta$0

This is contrasted with time-nonlocal memory-kernel equations of the form

$\eta$1

which do not admit such a simple semigroup structure (Paul et al., 13 May 2026).

Time locality does not, however, imply trace preservation. For $\eta$2, the map is completely positive and trace preserving, namely a standard GKSL semigroup. For $\eta$3, $\eta$4 decays, and the normalized evolution is not CP on the trace-preserving normalization, even though $\eta$5 remains a well-defined time-local exponential (Paul et al., 13 May 2026). In monitored realizations, the conditional state must therefore be continually renormalized.

A broader time-local setting appears in Quantum Liouvillian Tomography, where an open system satisfies

$\eta$6

with canonical form

$\eta$7

Here the rates $\eta$8 may become negative for $\eta$9; when this occurs, they signal CP-indivisibility and information back-flows characteristic of non-Markovian dynamics (Aguiar et al., 14 Apr 2025). Accordingly, time locality and Markovianity are distinct notions in the contemporary literature.

4. Continuous measurement and post-selection

The operational meaning of the hybrid Liouvillian is given by continuous measurement and selective retention of quantum trajectories. In the qubit Leggett–Garg derivation, each infinitesimal step $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$0 is modeled by coupling the system to a two-level ancilla, evolving under

$$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$1

and then projectively measuring the detector. This yields Kraus operators

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

Imperfect post-selection means that the $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$3 path is rescaled by $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$4, keeping only a fraction $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$5 of jump events; in the limit $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$6 this reproduces the hybrid master equation exactly (Paul et al., 13 May 2026).

In the detector-efficiency formulation, a detector of efficiency $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$7 registers a jump of type $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$8 with probability $$\frac{d\rho}{dt} = -\,i[H,\rho]-\gamma\{L^\dagger L,\rho\}+2\gamma q\,L\rho L^\dagger,$$9 and misses it with probability $q\in[0,1]$0, where

$q\in[0,1]$1

Post-selecting the no-click trajectory means imposing

$q\in[0,1]$2

The corresponding probability decays as

$q\in[0,1]$3

and the renormalized conditional mean evolution is generated by the hybrid Liouvillian (Minganti et al., 2020).

For intermediate efficiencies, the effect is nonlinear at the level of the normalized state. In the qubit analysis, $q\in[0,1]$4 means imperfect detectors record only a fraction $q\in[0,1]$5 of jumps, so the conditional state must be continually renormalized, producing a nonlinear back-action on the Bloch vector (Paul et al., 13 May 2026).

5. Spectral structure and exceptional points

One of the main uses of time-local hybrid Liouvillians is the interpolation between Hamiltonian and Liouvillian exceptional points. In the qubit Leggett–Garg model, vectorization in the basis

$q\in[0,1]$6

gives a $q\in[0,1]$7 superoperator. One eigenvalue is exactly $q\in[0,1]$8, while the remaining three satisfy

$q\in[0,1]$9

Exceptional points occur when the discriminant vanishes,

$$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$0

In the $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$1 Lindblad limit one recovers the standard purely real/negative spectrum, whereas in the $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$2 non-Hermitian limit the generator reduces to the no-jump $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$3 and the spectrum collapses onto $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$4 at $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$5, a third-order EP (Paul et al., 13 May 2026).

In red-sideband cavity optomechanics, the hybrid spectrum is obtained from a thermofield $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$6 matrix $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$7 whose characteristic polynomial factors as

$$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$8

The EP condition is $$\partial_t\rho = -\,i[H,\rho] -\frac12\sum_\mu\{L_\mu^\dagger L_\mu,\rho\} +\sum_\mu L_\mu\rho L_\mu^\dagger.$$9, which yields an explicit family $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$0. The limiting cases are

$$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$1

The first is independent of $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$2, while the second is thermally shifted by $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$3. Expanding about $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$4 gives

$$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$5

so the Hamiltonian EP is stable against weak quantum-jump perturbations to leading order (Ghosh et al., 25 Feb 2026).

In atomic vapors, the hybrid Liouvillian is represented as a full $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$6 matrix in Liouville space, with the only $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$7-dependence appearing in the last row and column, reflecting the fact that quantum jumps repopulate the diagonal identity component. The spectral consequences are substantial: for $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$8, $$H_{\rm eff}\equiv H-\frac{i}{2}\sum_\mu L_\mu^\dagger L_\mu$$9, there is an EP3 at $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$0 for $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$1, while for $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$2, $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$3, only an EP2 remains near $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$4. In the detuned case, the NHH superoperator shows a genuine ninefold EP at $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$5 and $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$6, whereas in the full Liouvillian spectrum only a threefold EP remains, at a slightly shifted $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$7 in the quoted numerical example (Kopciuch et al., 3 Jun 2025).

6. Physical consequences, applications, and interpretive cautions

In Leggett–Garg tests, the hybrid parameter directly controls the passage between standard Lindblad dynamics and near-algebraic temporal nonclassicality. The key result is that $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$8 approaches its algebraic maximum of $$\mathcal L_{\rm eff}[\rho] = -\,i(H_{\rm eff}\rho-\rho H_{\rm eff}^\dagger), \qquad \mathcal J[\rho] = \sum_\mu L_\mu\rho L_\mu^\dagger.$$9 in the null-efficiency limit, but even an infinitesimal increase in detector efficiency induces a rapid, highly nonlinear suppression toward the classical bound. The associated logarithmic sensitivity shows that maximal LGI violations are not robust physical features but singular limits of idealized measurement conditions. The same work emphasizes that achieving algebraic LGI violations in continuously evolving systems requires near-perfect suppression of detected quantum jumps, that is, effective post-selection (Paul et al., 13 May 2026).

In adiabatic control problems, hybrid-Liouvillian dynamics were used to design chiral state-conversion protocols with pure final states, no probability loss, and high fidelity. Extending beyond continuous adiabatic evolution, a stepwise protocol was constructed that facilitates conversion to pure states with fidelity $\chi$00 and, at the same time, no probability loss (Kumar et al., 2021). This establishes a concrete use of time-local hybrid interpolation: the non-Hermitian sector supplies pure-state eigenmodes, while the Lindbladian sector supplies probability conservation.

In exceptional-point physics more broadly, the formalism clarifies why a non-Hermitian Hamiltonian alone may be insufficient. For atomic vapors, the NHH approach alone may be insufficient to fully capture the system’s spectral properties, because quantum jumps can alter the existence, location in parameter space, or even the order of spectral degeneracies (Kopciuch et al., 3 Jun 2025). In optomechanics, the difference between the Liouvillian and Hamiltonian exceptional points provides a probe for thermal baths through the thermal shift of the conditional EP (Ghosh et al., 25 Feb 2026).

A related methodological development is Quantum Liouvillian Tomography, which introduces a protocol to capture and quantify non-Markovian effects in time-continuous quantum dynamics. The method reconstructs dynamical maps via gradient-based quantum process tomography and uses regression over the derivatives of Pauli string probability distributions to extract the Liouvillian governing the dynamics. It was applied to analyze the evolution of an idling two-qubit system implemented on a superconducting quantum platform and reported small but statistically significant negative regions in the dissipative rates, identified as signatures of non-Markovian backflow (Aguiar et al., 14 Apr 2025).

Two recurrent misconceptions are explicitly corrected by this body of work. First, time-locality does not imply trace preservation, complete positivity after normalization, or even Markovianity in the sense of CP-divisibility. Second, extreme spectral or temporal signatures obtained in the no-jump limit are not automatically features of the full open-system dynamics; quantum jumps can shift, split, lower, or remove them, and the hybrid Liouvillian is precisely the framework in which that passage is made explicit (Paul et al., 13 May 2026, Kopciuch et al., 3 Jun 2025).