Davies Equation in Open Quantum Systems

• The Davies Equation is a GKLS master equation that decomposes system–bath interactions by Bohr frequencies to ensure complete positivity and enforce detailed balance.
• It guarantees Gibbs state stationarity by linking jump rates to energy differences via Kubo–Martin–Schwinger conditions, thereby modeling thermalization.
• It serves as a tool for modeling quantum noise, resolving dissipation channel by channel in systems such as triple quantum dot spin qubits and extending to pseudodifferential operators.

The Davies equation, often called the Davies generator, is the weak-coupling, energy-resolved Gorini–Kossakowski–Lindblad–Sudarshan (GKLS) master equation obtained from a system weakly coupled to stationary thermal baths. Its defining structure is a decomposition of each system–bath coupling operator into components indexed by the Bohr frequencies of the system Hamiltonian, together with rates obeying a Kubo–Martin–Schwinger (KMS) detailed-balance relation. In that form, the dynamics is completely positive and trace preserving, enforces detailed balance, and has the Gibbs state as a fixed point. In the literature considered here, the Davies equation appears both as the canonical thermalizing Lindbladian and as a concrete analytic tool for noise modeling in triple quantum dot spin qubits, for comparisons between quantum and embedded classical Markov dynamics, and for localized constructions that extend the framework to unbounded and pseudodifferential operators (Mehl et al., 2012, Shiraishi et al., 14 Jul 2025, Basso et al., 8 Oct 2025, Galkowski et al., 31 Mar 2026).

1. Weak-coupling construction and canonical GKLS form

In its standard form, the Davies equation is built from a system Hamiltonian $H_S$ and coupling operators $S_\alpha$ by first resolving each $S_\alpha$ into Bohr-frequency components,

$$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$

where $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$ and $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$. The corresponding GKLS generator is

$$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$

with rates determined by bath correlation spectra and a Lamb shift given by the principal-value part of the same bath data (Shiraishi et al., 14 Jul 2025).

The triple-quantum-dot analysis adopts the dissipative part in an explicitly energy-resolved form,

$$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$

with

$$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$

Here $\mathcal A_\omega$ acts only between system eigenstates whose energy difference equals $S_\alpha$0. This “energy-resolving” of jump operators is the operational core of the Davies construction: the dissipator is organized channel by channel according to the system’s Bohr spectrum rather than by ad hoc phenomenological decay terms (Mehl et al., 2012).

A common source of confusion is to identify the Davies equation with an arbitrary Markovian thermal Lindbladian. In the Davies construction, the frequency decomposition is not incidental. It is the mechanism that ties each jump process to the spectral structure of $S_\alpha$1, and it is this structure that underlies both thermodynamic consistency and the separation of dephasing, relaxation, and leakage channels in concrete device models (Mehl et al., 2012, Shiraishi et al., 14 Jul 2025).

2. Detailed balance, Gibbs stationarity, and thermodynamic structure

The thermodynamic content of the Davies equation is encoded in its rate symmetry. In the triple-dot notation, the rates satisfy

$S_\alpha$2

while in the standard inverse-temperature notation one has

$S_\alpha$3

These are KMS or detailed-balance relations. They imply that the Gibbs state is stationary: $S_\alpha$4 In the observable-picture formulation, the Davies generator is reversible in the KMS inner product

$S_\alpha$5

which yields a Dirichlet form and a well-defined spectral-gap variational principle (Mehl et al., 2012, Basso et al., 8 Oct 2025).

This combination of detailed balance, Gibbs stationarity, and complete positivity is the principal reason the Davies equation is used as a reference thermalizing model. The 2012 triple-dot work introduces it “to make sure that the system equilibrates in the long time limit,” while the 2025 quadratic-systems work emphasizes that the resulting generator is completely positive and trace preserving and has the Gibbs steady state (Mehl et al., 2012, Shiraishi et al., 14 Jul 2025).

The same structure clarifies what is and is not included. In the triple-dot treatment, the Lamb-shift Hamiltonian is not tracked explicitly; only the dissipative Davies generator $S_\alpha$6 is retained, and coherent dynamics is removed by moving to a rotating frame. In that frame, the dissipator remains unchanged because the energy-resolved phases cancel pairwise inside $S_\alpha$7 (Mehl et al., 2012).

3. Secular approximation, locality, and the non-secular quadratic result

In the standard presentation, the Davies equation is associated with the weak-coupling or van Hove limit together with an energy-resolved secular approximation. In the triple-dot formulation, the decomposition $S_\alpha$8 and the frequency-selection rule implement that secular approximation automatically. This is precisely why the model avoids the positivity and detailed-balance problems that can arise in non-secular, Redfield-type generators (Mehl et al., 2012).

A more recent result changes the scope of that statement for quadratic systems. For number-conserving quadratic fermion or boson systems coupled linearly to independent, identical baths at each site, the quasi-local Redfield equation coincides exactly with the Davies equation without invoking the secular approximation. In that setting, the $S_\alpha$9 coherence-generating terms cancel algebraically after summing over sites because

$S_\alpha$0

with $S_\alpha$1 the unitary transformation between site operators and normal modes. The resulting generator is the Davies GKLS generator, with jump operators $S_\alpha$2 and $S_\alpha$3, detailed balance, Gibbs steady state, and complete positivity, even when vanishing energy-level spacings would ordinarily invalidate secular coarse-graining (Shiraishi et al., 14 Jul 2025).

This result directly addresses a persistent tension between locality and thermodynamic consistency. Standard secular Davies generators are global in the energy eigenbasis, while local Redfield generators are microscopically motivated but may violate complete positivity or detailed balance. For the quadratic models just described, that opposition disappears: the quasi-local Redfield equation equals the Davies generator exactly. The paper is explicit, however, that the mechanism depends on linear, number-conserving couplings, identical independent baths, and the unitarity identity above; for interacting or nonlinear systems, the same cancellation generally fails (Shiraishi et al., 14 Jul 2025).

A plausible implication is that the secular approximation is not the essential ingredient in every rigorous route to Davies dynamics. In the quadratic setting, algebraic cancellation replaces dynamical averaging, so degeneracies or dense spectra do not by themselves obstruct a thermodynamically consistent GKLS description (Shiraishi et al., 14 Jul 2025).

4. Triple quantum dot spin qubits and frequency-resolved noise modeling

In the triple quantum dot spin system, the Hamiltonian is

$S_\alpha$4

with exchange couplings $S_\alpha$5 tunable via bias $S_\alpha$6 and Zeeman energy $S_\alpha$7. The spectrum contains four spin-$S_\alpha$8 quadruplets $S_\alpha$9 and four spin-$$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$0 doublets $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$1. The Bohr frequencies entering the Davies decomposition are the energy differences between these eigenstates (Mehl et al., 2012).

Two encoded qubits are considered. The subspace qubit uses the computational subspace $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$2, with logical “1” identified with $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$3 and logical “0” with $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$4. The subsystem qubit uses the four doublet states $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$5 and encodes information in the $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$6 label, while the Zeeman-split $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$7 degree of freedom is treated as an irrelevant subsystem held at the thermal state

$$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$8

The corresponding projection maps define the encoded qubit state used in the reduced dynamics (Mehl et al., 2012).

The environment model is formulated directly in Davies form. The dominant operators are local $$A_\alpha(\omega)=\sum_{\varepsilon'-\varepsilon=\omega}\Pi_\varepsilon S_\alpha \Pi_{\varepsilon'},$$9 terms for dephasing and local $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$0 terms for spin relaxation. For hyperfine noise, the rates are Gaussian in $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$1 with detailed-balance factors,

$H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$2

with typical $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$3 and $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$4. For phonon-mediated relaxation,

$H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$5

with $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$6. These rates are not derived from bath integrals in the paper; instead they are parameterized directly as realistic, frequency-resolved functions $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$7 that obey KMS (Mehl et al., 2012).

5. Initial-time reductions, toy models, and encoded-qubit error channels

The triple-dot analysis separates an initial operation window $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$8 with $H_S=\sum_\varepsilon \varepsilon\,\Pi_\varepsilon$9 from microsecond-scale equilibration. Short-time encoded dynamics is derived by a Nakajima–Zwanzig projection-operator method, using second-order Born approximations after projecting to the encoded subspace or subsystem. The resulting reduced equations show that the initial time evolution relevant for qubit operation decouples from the long-time thermalization dynamics (Mehl et al., 2012).

Errors are parameterized as leakage, relaxation, and dephasing rates extracted at three Bloch-sphere points, $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$0, $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$1, and $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$2. The notation

$[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$3

is used to capture initial non-exponential behavior over the 10 ns window. Leakage is defined from the occupation loss, dephasing from the renormalized equatorial Bloch component $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$4, and entanglement fidelity is expanded explicitly in terms of the $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$5 (Mehl et al., 2012).

Because the Davies generator sorts transitions by $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$6, the observed errors are reproduced by four compact toy GKLS models: pure relaxation, pure dephasing, two-state leakage to an “Out” level, and internal thermally biased transitions within the subsystem qubit. In the pure-dephasing model, $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$7, so dephasing cancels when $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$8. In the triple-dot subspace qubit this produces a bias-dependent cancellation point at $[H_S,A_\alpha(\omega)]=-\omega A_\alpha(\omega)$9, approximately $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$0 (Mehl et al., 2012).

The regime analysis is explicit. Regime 1 is local phase noise $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$1, strong at small $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$2. Regime 2 is phonon-induced local spin relaxation $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$3, strong at large $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$4 and large $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$5. Regime 3 is hyperfine-induced local spin relaxation $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$6, dominant at small $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$7 and independent of $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$8. In GaAs, both subspace and subsystem encodings have similar coherence properties over 10 ns: local hyperfine phase noise is the principal limitation for both; hyperfine-induced relaxation matters only near crossings; phonon relaxation is minor on nanosecond scales; and avoiding crossings during operations minimizes leakage and relaxation. An additional asymmetry arises for noise on dot 1 and positive detuning, where errors are strongly suppressed because $$\dot{\rho} = -i[H_S+H_{\mathrm{LS}},\rho] + \sum_{\omega,\alpha}\gamma_\alpha(\omega) \left( A_\alpha(\omega)\rho A_\alpha^\dagger(\omega) -\tfrac12\{A_\alpha^\dagger(\omega)A_\alpha(\omega),\rho\} \right),$$9 approaches a double-dot singlet on dots 2–3 that is insensitive to noise on dot 1 (Mehl et al., 2012).

6. Embedded classical chains and spectral-gap comparison

For a purely dissipative Davies generator acting on observables,

$$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$0

the Bohr-frequency subspaces

$$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$1

are invariant, and the global spectral gap satisfies

$$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$2

The block $$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$3 consists exactly of observables commuting with $$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$4, and on that block the Davies dynamics contains an embedded classical Markov generator (Basso et al., 8 Oct 2025).

Given an eigenbasis $$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$5 of $$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$6, the associated classical chain has stationary distribution $$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$7 and transition rates

$$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$8

The paper proves that there exists an eigenbasis $$\widetilde{\mathcal{L}_D(\rho)}=\sum_{\mathcal A,\omega} h(\mathcal A,\omega)\,\mathcal D[\mathcal A_\omega](\rho),$$9 such that

$$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$0

and $$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$1 minimizes the corresponding classical spectral gap over eigenbases. Thus the commuting sector of Davies dynamics is not merely analogous to a classical reversible chain; it is precisely identified with one after the appropriate basis choice (Basso et al., 8 Oct 2025).

The main comparison theorem isolates long arithmetic progressions in $$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$2 as the obstruction to quantum–classical gap comparability. If $$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$3 contains no proper $$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$4-term arithmetic progression, then

$$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$5

Equivalently, for the minimizing basis $$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$6,

$$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$7

If the spectrum is simple, this becomes

$$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$8

The paper further shows that for a large class of Hamiltonians, including generic local-field perturbations, $$\mathcal D[\mathcal A](B)\equiv \mathcal A B\mathcal A^\dagger-\tfrac12(\mathcal A^\dagger\mathcal A B+B\mathcal A^\dagger\mathcal A), \qquad \mathcal A=\sum_\omega \mathcal A_\omega.$$9 almost surely, so the quantum gap remains within a constant factor of the classical gap (Basso et al., 8 Oct 2025).

This result sharpens the interpretation of off-diagonal decay. Absent long arithmetic progressions, decoherence in $\mathcal A_\omega$0 sectors is not parametrically slower than relaxation of diagonal observables. The paper formalizes that statement through a reduction from off-diagonal eigenvectors to commuting observables and through gap bounds on the corresponding Rayleigh quotients (Basso et al., 8 Oct 2025).

7. Localized Davies generators and unbounded operators

The standard Davies construction requires precise knowledge of the Bohr spectrum, equivalently the spectral projections of the Hamiltonian. A time-localized variant replaces exact frequency filtering by convolution in time. For a self-adjoint Hamiltonian $\mathcal A_\omega$1, coupling operators $\mathcal A_\omega$2, and a window $\mathcal A_\omega$3, the filtered operator is

$\mathcal A_\omega$4

and the localized Lindbladian is

$\mathcal A_\omega$5

$\mathcal A_\omega$6

For Gaussian windows $\mathcal A_\omega$7, the limit $\mathcal A_\omega$8 recovers the exact spectral filtering, and in finite dimensions the localized generator converges to the Davies generator in Schatten norms; the coherent correction $\mathcal A_\omega$9 vanishes in norm in the same limit (Galkowski et al., 31 Mar 2026).

The extension to unbounded operators uses graph Sobolev spaces $S_\alpha$00 and commutator bounds ensuring dissipativity of

$S_\alpha$01

Under the stated square-summability and commutator assumptions, $S_\alpha$02 is a contraction on the trace class, preserves the trace, and is completely positive. For Gaussian $S_\alpha$03, exact Gibbs stationarity is recovered if the rate function satisfies the localized balance condition

$S_\alpha$04

together with the explicit coherent correction $S_\alpha$05 built from the kernels $S_\alpha$06 and $S_\alpha$07; then

$S_\alpha$08

In the delocalized limit $S_\alpha$09, this reduces to the standard KMS relation $S_\alpha$10 (Galkowski et al., 31 Mar 2026).

The pseudodifferential extension covers Schrödinger operators $S_\alpha$11 with confining $S_\alpha$12, self-adjoint elliptic operators on compact manifolds, and more general Weyl-quantized Hamiltonians $S_\alpha$13 under symbol-class assumptions. The significance is not merely technical. Time localization dispenses with the need to compute Bohr frequencies explicitly and constructs “quantum Gibbs samplers” from dynamical information $S_\alpha$14 alone, while preserving the central Davies feature of Gibbs stationarity under the localized balance condition (Galkowski et al., 31 Mar 2026).