Quantum Informational Dissipation

• Quantum informational dissipation is the process by which quantum information is degraded, redistributed, or purposefully routed via open-system dynamics, reducing fidelity and coherence.
• Key observables such as gate fidelity reduction, entropy production, and mutual information decay quantify the delocalization and loss of local accessibility of quantum information.
• Engineered dissipation leverages mechanisms like dark state preparation and autonomous state transfer to stabilize quantum states, aid error correction, and enable controlled measurement.

Quantum informational dissipation denotes a family of phenomena in which quantum information is degraded, redistributed, or deliberately routed by open-system dynamics. In weakly dissipative quantum operations it appears as a reduction in average gate fidelity that is linear in the gate time and dissipation rates at first order (Abad et al., 2021). In many-body dynamics it appears as decay of mutual information, delocalization of initially localized states, or loss of locally accessible information under boundary noise, random couplings, or generic linear dissipation (Alba et al., 2021, Farooq et al., 2014, Zhang et al., 2018, Lovas et al., 2023). In quantum thermodynamics it appears as entropy production, loss of system-reference correlations, and Landauer heat costs (Zhang et al., 2021, Riechers et al., 2020, Vu et al., 2021, Hashimoto et al., 2022). In engineered settings it is a resource for measurement, state preparation, stabilization, autonomous state transfer, and logical-channel synthesis (Harrington et al., 2022, Marshall et al., 2016, Wang et al., 2018, Dambal et al., 28 May 2026). This suggests that the expression functions less as a single invariant than as a unifying description of how dissipation changes the accessibility, distinguishability, and encoding of quantum information.

1. Formal settings and core definitions

A standard formal setting is the Markovian open quantum system described by a Lindblad master equation,

$$\partial_t \hat\rho = \frac{1}{i}[\hat H,\hat\rho] + \sum_\alpha \Gamma_\alpha \left( \hat L_\alpha \hat \rho \hat L_\alpha^\dagger -\frac{1}{2} \hat L_\alpha^\dagger \hat L_\alpha \hat \rho -\frac{1}{2} \hat \rho \hat L_\alpha^\dagger \hat L_\alpha \right),$$

with jump operators $\hat L_\alpha$ and rates $\Gamma_\alpha$ (Harrington et al., 2022). In this framework, dissipation engineering amounts to controlling the jump operators, and dark states satisfy $\hat L_\alpha\ket{\psi}=0$ for all $\alpha$; more generally, dark subspaces are null spaces of $$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$ (Harrington et al., 2022). The same review treats measurement as a dissipative interaction in which information is transferred from the system into external degrees of freedom, and continuous monitoring is represented by Kraus operators whose average reproduces Lindblad dynamics (Harrington et al., 2022).

A distinct but related usage arises when the microscopic evolution is fully unitary yet dissipation is informational and effective. In random quantum networks of spin-$1/2$ particles with randomly changing $XY$ couplings, each realization evolves unitarily, but the ensemble-averaged reduced state is mixed; the resulting “information dissipation” is the loss of local access to initially stored single-qubit information or two-qubit entanglement as it becomes delocalized over the network (Farooq et al., 2014). This distinction matters because several works use the same language of dissipation for both genuine open-system decoherence and effective information loss caused by averaging and partial tracing (Farooq et al., 2014, Harrington et al., 2022).

A thermodynamic formulation places the emphasis on entropy production. For an arbitrary initial joint state of system and environment, thermodynamic dissipation can be written as a sum of information-theoretic quantities,

$\frac{1}{k_B}\EP = \Delta \mathcal{I}(\rho_t ; \rho_t^{\text{env}}) + \Delta D\bigl[\rho_t^{\text{env}}\big\| \sigma^{\text{env}} \bigr],$

so dissipation is identified with the increase of system-environment correlations and the increase of the environment’s informational distance to a reference state (Riechers et al., 2020). In that sense, quantum informational dissipation is not only loss of coherence or purity but also a precisely quantifiable redistribution of information into correlations and environmental nonequilibrium (Riechers et al., 2020).

2. Operational observables: fidelity, entropy, and mutual information

At the level of quantum information processing, weak dissipation produces a universal first-order reduction of average gate fidelity. For an $N$-qubit Lindblad dynamics with $\hat L_\alpha$0, the average gate fidelity satisfies

$\hat L_\alpha$1

and the first-order reduction is independent of the specific unitary gate: it depends only on the gate time and the dissipation (Abad et al., 2021). The same work shows that a large class of additive collective noise processes, including $\hat L_\alpha$2 and $\hat L_\alpha$3, gives the same average gate fidelity reduction as uncorrelated noise of similar strength, whereas non-additive processes such as two-photon relaxation $\hat L_\alpha$4 produce additional fidelity loss detectable in two-qubit gates (Abad et al., 2021). A direct implication stated there is that average gate fidelity alone is a poor diagnostic for spatial correlations relevant to quantum error correction (Abad et al., 2021).

Entropy-like observables provide a complementary view. In transverse-field Ising chains subject to generic linear dissipation, entropy-related quantities such as the von Neumann entropy, the Rényi entropies, and the associated mutual information admit a hydrodynamic description within a quasiparticle picture in the limit of long times, large system sizes, and weak dissipation (Alba et al., 2021). A universal feature is that the mutual information grows up to a time scale proportional to the inverse dissipation rate and then decreases, always vanishing in the long-time limit (Alba et al., 2021). In random quantum networks, average fidelity, linearized von Neumann entropy, and concurrence play the same diagnostic role: decay of local fidelity or concurrence, and growth of linear entropy, quantify the loss of local or pairwise information as it spreads into the rest of the network (Farooq et al., 2014).

These observables emphasize different facets of dissipation. Fidelity reduction measures deviation from a target operation (Abad et al., 2021). Mutual information and entropies measure loss or redistribution of correlations (Alba et al., 2021). Linear entropy and concurrence quantify the degradation of local purity and bipartite entanglement under either open or effectively dissipative dynamics (Farooq et al., 2014).

3. Many-body spreading, hydrodynamics, and coding transitions

A central many-body theme is the competition between scrambling and leakage. In dissipative Ising chains, the hydrodynamic quasiparticle picture yields analytic formulas for arbitrary functions of the fermionic covariance matrix and predicts that mutual information eventually vanishes for any possible dissipation (Alba et al., 2021). In random quantum networks, the weighting parameter $\hat L_\alpha$5 controls the typical connectivity of the network and therefore the rate of information delocalization: $\hat L_\alpha$6 produces almost no dynamics, intermediate $\hat L_\alpha$7 yields damped loss of local information, and $\hat L_\alpha$8 produces recurrent, nearly coherent oscillatory behavior with revivals (Farooq et al., 2014).

Boundary loss leads to a sharper notion of a coding transition. In a one-dimensional qudit chain evolved by local Haar-random unitaries with depolarizing noise at one boundary, locally accessible quantum information remains partially protected when the dissipation is sufficiently weak, and up to time scales growing linearly in system size $\hat L_\alpha$9; for strong enough dissipation, that information is completely lost to the dissipative environment (Lovas et al., 2023). The same work shows that scrambling the information with a unitary circuit of depth $\Gamma_\alpha$0 before the onset of dissipation can perfectly protect the information until late times, and that at weak dissipation it is possible to code at a finite rate, so that a fraction of the many-body Hilbert space can be used to encode quantum information (Lovas et al., 2023).

This body of results supports a precise distinction between mere spreading and coding. Scrambling alone redistributes information nonlocally; weak boundary dissipation can coexist with a protected encoded sector for times of order $\Gamma_\alpha$1; sufficiently strong dissipation destroys that sector; and suitable preshuffling of information can improve protection dramatically (Lovas et al., 2023). The hydrodynamic and random-network results show the same competition in less coding-oriented language: information is not destroyed globally in the closed-system description, but it becomes inaccessible to local probes or reduced subsystems (Farooq et al., 2014, Alba et al., 2021).

4. Scrambling, dissipative locality, and random quantum networks

Dissipative scrambling requires separate treatment because raw correlators mix leakage and operator growth. In a chaotic Ising chain with Lindblad dissipation, a dissipative version of the out-of-time-ordered correlator is defined from an interferometric protocol with forward and backward open-system evolution (Zhang et al., 2018). The raw dissipative OTOC decays rapidly for all sites because information leaks to the environment, so the light-cone structure is obscured. After dividing by a leakage-only factor $\Gamma_\alpha$2, the corrected OTOC reveals an approximately ballistic light cone at intermediate scales, but the cone reaches only a finite distance: even after removing the overall decay, dissipation changes the structure of operator spreading so that the information light cone becomes spatially finite (Zhang et al., 2018).

The same work conjectures a normalized Lieb-Robinson-type bound,

$\Gamma_\alpha$3

and numerically finds that the characteristic width of the recoverable light cone obeys a power law $\Gamma_\alpha$4; for phase damping and phase depolarizing, fitted exponents lie near $\Gamma_\alpha$5–$\Gamma_\alpha$6, while an analytical lower bound gives $\Gamma_\alpha$7 for sufficiently small $\Gamma_\alpha$8 (Zhang et al., 2018). The mechanism is that dissipation suppresses many-body operator strings more strongly than few-body strings, so late-time operator weight is dominated by one-body and nearest-neighbor two-body terms (Zhang et al., 2018).

Random quantum networks furnish a complementary picture in which the network topology itself fluctuates. There the excitation initially localized at one site disperses through a sequence of random adjacency matrices, and the ensemble-averaged reduced state exhibits decay of fidelity and concurrence together with growth of linear entropy (Farooq et al., 2014). The work explicitly stresses that this dissipation is not thermodynamic energy dissipation: it is effective informational dissipation generated by averaging over random graph realizations and by tracing out most of the system (Farooq et al., 2014). Together with the dissipative OTOC results, this shows that “information dissipation” can denote both genuine environmental leakage and emergent irreversibility of locally accessible information under complex unitary many-body dynamics (Farooq et al., 2014, Zhang et al., 2018).

5. Engineered, autonomous, and programmable dissipation

A major modern theme is that dissipation can be engineered rather than merely tolerated. Controlled dissipation is presented as an essential tool for quantum measurement, quantum state preparation, and quantum state stabilization, and the same review emphasizes its role in quantum error correction and quantum simulation (Harrington et al., 2022). Within the Lindblad framework, dark states and decoherence-free subspaces are selected by the jump operators, while strong dissipation realizes Zeno dynamics that confines the system to the kernel of $\Gamma_\alpha$9 (Harrington et al., 2022). Concrete examples include optical pumping to a dark state, dissipative Bell-state preparation, Raman sideband cooling, and bosonic code stabilization; cat states, for example, can be stabilized with a two-photon jump operator

$\hat L_\alpha\ket{\psi}=0$0

for which cat states are approximate dark states (Harrington et al., 2022).

Several works push this resource-theoretic viewpoint further. A modular model based on fast dissipation uses a strong Liouvillian to create a steady-state subspace and weak controls to induce effective Hamiltonian or Lindbladian dynamics within it; the resulting dissipation-generated modules can prepare all single-qubit gates and the CNOT gate, simulate arbitrary Lindbladian dynamics, and realize a quantum memory in a dissipative environment with tunable loss of coherence and concurrence (Marshall et al., 2016). Autonomous quantum state transfer by dissipation engineering shows that a qubit of information can be transferred directionally and without time-dependent control by a dissipative process; the minimum system dimension for transferring one qubit of information is $\hat L_\alpha\ket{\psi}=0$1, plus one auxiliary reservoir, and the effective jump operator transfers amplitudes from the sender to the receiver while preserving the logical coefficients (Wang et al., 2018).

A logical-channel formulation appears in programmable dissipation via partial quantum error correction. There one fault-tolerant round induces a logical CPTP map, and decoder or recovery randomization generates a controllable family of logical channels whose convex mixtures realize Kraus-channel mixing (Dambal et al., 28 May 2026). This enables direct compilation of target dissipators into effective logical dynamics without explicit ancilla qubits for encoding bath degrees of freedom, and the required code distance is chosen so that uncontrolled logical errors remain a small fraction of the intended dissipation per step, rather than being driven below an arbitrarily small closed-system tolerance (Dambal et al., 28 May 2026). This suggests an overview between fault tolerance and reservoir engineering: the error-correction cycle itself becomes a programmable dissipative primitive (Dambal et al., 28 May 2026).

6. Reference systems, entropy production, and fluctuation theorems

An explicitly informational thermodynamic formulation introduces a reference system $\hat L_\alpha\ket{\psi}=0$2 initially correlated with the system $\hat L_\alpha\ket{\psi}=0$3 but not with the environment $\hat L_\alpha\ket{\psi}=0$4. In that setting, the conditional entropy production with respect to the reference exceeds the ordinary entropy production, and their difference is the dissipative information

$\hat L_\alpha\ket{\psi}=0$5

which quantifies the loss of $\hat L_\alpha\ket{\psi}=0$6–$\hat L_\alpha\ket{\psi}=0$7 correlations and the creation of distributive correlations between $\hat L_\alpha\ket{\psi}=0$8 and $\hat L_\alpha\ket{\psi}=0$9 even though they do not interact directly (Zhang et al., 2021). The same work shows that a process may have zero unconditional entropy production at thermal equilibrium while still having nonzero conditional entropy production with respect to the reference, and that $\alpha$0 is a minimal thermodynamic cost associated with maintaining and then losing those correlations (Zhang et al., 2021). Both entropy productions and the dissipative information satisfy fluctuation theorems, and the fluctuation theorem for dissipative information was verified experimentally on IBM quantum computers (Zhang et al., 2021).

A second exact result identifies the initial-state dependence of thermodynamic dissipation with contraction of relative entropy to a minimally dissipative reference state. For any process with fixed initial environment,

$\alpha$1

so the additional dissipation incurred by using $\alpha$2 rather than the minimally dissipative state $\alpha$3 is exactly the lost distinguishability between their trajectories (Riechers et al., 2020). The same work decomposes this into a classical Kullback-Leibler contraction plus a quantum coherence term $\alpha$4, and shows that for nonunitary state preparation mismatched expectations can lead to divergent dissipation as the actual initial state becomes orthogonal to the anticipated one (Riechers et al., 2020).

A closely related fluctuation-theorem approach addresses the spread of quantum information itself. For a tripartite state $\alpha$5 evolved by a local unitary on $\alpha$6, the decrease of mutual information,

$\alpha$7

is treated as informational dissipation (Zhang et al., 2022). The underlying stochastic quantity is a change in stochastic quantum mutual information $\alpha$8, and the exact quasiprobability fluctuation theorem

$\alpha$9

holds for a normalized but generally nonpositive quasiprobability distribution $$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$0 (Zhang et al., 2022). The need for quasiprobabilities rather than genuine probabilities is traced to noncommutativity between global and local projectors, and the relation was also tested on an IBM quantum processor (Zhang et al., 2022).

7. Landauer bounds, coherence costs, and newer frontier phenomena

Quantum informational dissipation also appears in finite-time information erasure. For a Markovian open quantum system satisfying detailed balance, the dissipated heat obeys the strengthened finite-time Landauer bound

$$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$1

together with a coherence-based lower bound

$$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$2

where $$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$3 is the coherence accumulated in the instantaneous energy basis (Vu et al., 2021). The paper concludes that creation of quantum coherence in the energy eigenbasis during erasure inevitably leads to additional heat costs, and its optimal-control example finds that the minimal-dissipation protocol avoids coherence generation and behaves essentially classically in the energy basis (Vu et al., 2021). A complementary study of a qubit coupled to a bosonic reservoir with a tilted interaction direction finds that the entropic Landauer bound is tighter for a sufficiently mixed initial state, while the thermodynamic fluctuation bound is tighter when the purity of the initial state is sufficiently high; in the pure-dephasing case the two bounds coincide when the initial coherence is zero, otherwise the thermodynamic bound is tighter (Hashimoto et al., 2022). The same work argues that information erasure inevitably accompanies constant energy dissipation caused by the creation of system-reservoir correlation (Hashimoto et al., 2022).

Two recent directions widen the scope further. One derives a universal bound on the efficiency with which dissipated work generates distinguishable changes in a finite-temperature quantum many-body state, quantified by the Bures quantum Fisher information. The frequency-resolved efficiency kernel

$$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$4

is bounded by $$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$5, which yields

$$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$6

with a crossover at the Planckian scale $$\hat Q=\sum_\alpha \hat L_\alpha^\dagger \hat L_\alpha$$7; the same work states that Planckian scatterers sit at the edge of optimality and that the bounded quantity can be evaluated directly from optical conductivity measurements (Chowdhury, 4 Feb 2026). Another studies driven non-Hermitian qubit systems with collective reservoir engineering and finds an informational Mpemba effect: a more mixed initial state reaches its steady state with unit purity at a faster rate, and the onset of efficient purification-assisted entanglement generation is dictated by the degeneracy of collective subradiant modes rather than by exceptional points (Feyisa et al., 9 Apr 2026). This suggests that dissipation can sometimes accelerate purification and entanglement generation rather than merely degrading them (Feyisa et al., 9 Apr 2026).

Across these formulations, quantum informational dissipation consistently names the irreversible reorganization of quantum information under dynamics that are noisy, monitored, coarse-grained, or deliberately engineered. What changes from one subfield to another is the preferred observable—gate fidelity, mutual information, Rényi entropy, OTOC, relative entropy, heat, or quantum Fisher information—and the operational question attached to it: preservation, diagnosis, control, or exploitation of dissipation (Abad et al., 2021, Alba et al., 2021, Zhang et al., 2018, Harrington et al., 2022, Zhang et al., 2021, Chowdhury, 4 Feb 2026).