Skip to content

Operational quasiprobability in quantum thermodynamics - Work extraction by coherence and nonjoint measurability

FieldValue
JournalPhysical Review Research
Published2026-02-06
AuthorsJeongwoo Jae, Junghee Ryu, Hoon Ryu
DOI10.1103/qbs6-nj3l

Operational quasiprobability work exceeded the jointly measurable bound for a sharp qubit measurement, with μ=1. A modeled three-level nitrogen-vacancy example at Ω1=Ω2=4.4π MHz yielded identical extractable work from OQ and MHQ despite different quasiprobability negativities.

A work distribution built from two settings

Section titled “A work distribution built from two settings”

The operational quasiprobability, or OQ, assigns a value q_if to each initial and final measurement outcome. Its construction combines an end-point measurement, EPM, at t2, with a two-point measurement, TPM, at t1<t2. The TPM measures the initial energy A, evolves the state, and then measures the final energy B_H. The EPM measures B at the endpoint without the preceding measurement.

The OQ follows from an inverse Fourier transform of the characteristic functions for these settings. Its marginals reproduce the separate measurement probabilities for A and B. Unlike the Wigner function, its representation does not depend on the functional form of the observables.

Three properties support its use as a work distribution: marginality, TPM reproducibility, and convex linearity. For an input diagonal in A’s basis, the OQ equals the TPM probability. For a convex mixture of states, the OQ is the corresponding weighted mixture of their values. These properties retain measured marginals while accounting for the state disturbance introduced by sequential measurements.

The TPM scheme reproduces the Jarzynski equality for a Gibbs state under arbitrary unitary evolution. However, its first measurement disturbs the state and changes the average from the energy-difference definition. The OQ expectation instead gives ⟨w⟩OQ=⟨Ef⟩B−⟨Ei⟩A, matching the classical average. Each outcome carries work w=Ef−Ei, including negative excitation contributions when the distribution has negative weights.

No positive joint distribution can preserve both measurement marginals and satisfy the Jarzynski equality in this setting. Negative OQ values expose the mismatch that a classical work distribution cannot represent.

The Jarzynski equality links the exponential average of work to the free-energy difference between equilibrium states. Coherence produces a correction to this relation and to the work moments.

For an incoherent input, the OQ and TPM corrections agree because ΓOQ=ΓTPM. For a coherent input, the modified Jarzynski relation takes the form ⟨e^−βw⟩OQ=e^−βΔFΓOQ. The OQ correction uses the overlap between the initial coherent state and the final Gibbs state. The TPM correction measures only the incoherent part because the initial measurement removes coherence.

The work difference is δw=⟨w⟩OQ−⟨w⟩TPM. It satisfies δw≥δF, where δF=F(rho)−F(rhoD) compares the coherent state with its incoherent counterpart. The coherence term is Crel=S(rhoD)−S(rho), which is nonnegative. Equality holds only when rho is incoherent. The bound can also be set using entrywise 1-norm coherence and the off-diagonal Hamiltonian norm. This identifies initial coherence as a source of average work that the TPM scheme omits.

The OQ second moment can be negative, unlike the quasiprobabilities listed in reference 6. The paper still provides a lower bound tied to the coherent part of the initial state. That bound assumes a symmetric energy profile with the energy eigenvalues summing to zero. The appendix proof also treats A and B as projectors. The second-moment result therefore has narrower conditions than the general OQ work distribution.

Qubit work crosses the joint measurability bound

Section titled “Qubit work crosses the joint measurability bound”

Work extraction is analyzed with generalized positive operator-valued measures, or POVMs, at both protocol times. For binary unbiased qubit measurements, each POVM element is μΠ+(1−μ)I/2, where μ controls sharpness. μ=1 gives projective elements, while μ=0 gives I/2 for each outcome.

Two measurements are jointly measurable when one joint POVM can reproduce both sets of marginals. The paper proves that OQ is nonnegative for every qubit state exactly when these binary unbiased measurements are jointly measurable. For unbiased measurements, this condition matches Busch’s criterion: ||vi+vf||+||vi−vf||≤2. Non-joint measurability is the complementary case.

For a positive OQ, the authors derive a state-independent classical work bound Wcl. They optimize this bound over measurement sharpness to define the jointly measurable limit max_μWcl. The maximizing Bloch vector for the positive-OQ bound is r_max=−(vi−vf)/||vi−vf||.

The authors define extractable work as the energy remaining at the end of the protocol. Positive de-excitation weights increase it, while excitation weights reduce it. Negative excitation weights therefore increase the OQ result.

The qubit calculation uses p=1/2, c=1/2, δ=(√2+1)Ω, and Ω=1. It also uses equal sharpness for the initial and final measurements. The outcome (0,1) has a negative OQ value and represents excitation. Figure 2 shows sharp non-jointly measurable measurements exceeding the jointly measurable bound. The TPM scheme gives zero extractable work in the same comparison. The theorem provides an operational certificate: work above max_μWcl implies non-joint measurability.

The Kirkwood-Dirac quasiprobability, KDQ, can contain imaginary values for two POVMs. Its real part is the Margenau-Hill quasiprobability, MHQ. MHQ reconstruction requires EPM, TPM, and weak-TPM data. Weak-TPM first applies the initial measurement without selecting an outcome, then measures the final energy. OQ needs only EPM and TPM data.

For any binary measurement on a two-dimensional system, the paper proves that OQ equals MHQ. This extends an earlier projective-measurement equivalence to general, possibly biased binary measurements. For unbiased qubit measurements, MHQ is nonnegative for all states exactly when A and B_H are jointly measurable.

KDQ negativity is a stricter criterion than noncommutativity. MHQ negativity identifies the joint-measurability violation in this setting. Two-level experiments can therefore use OQ instead of the three-setting MHQ reconstruction.

The three-level calculation models the spin-triplet nitrogen-vacancy center used in prior work. A bichromatic microwave field drives the transitions from |0⟩ to |−1⟩ and from |0⟩ to |1⟩. Here Ω1 and Ω2 are Rabi frequencies, while φ1 and φ2 set the rotating-frame phases.

The chosen parameters are Ω1=Ω2=4.4π MHz and φ=1.09Ω. Projective measurements use eigenvectors of the effective Hamiltonian. The initial pure state minimizes the MHQ excitation value from |−1⟩ to |1⟩.

The paper specifies the initial state through these values.

Level ip_ia_i
10.76540.0073
00.00090.2787
−10.23380.0002

The preparation uses a pure state with the tabulated p_i values and phase parameters a_i. Here |0⟩ becomes a dark state under stimulated Raman adiabatic passage in the resonant condition. The zero EPM probability for |0⟩ reflects that dark state.

OQ and MHQ do not coincide in this three-level system. Their negativity measures sum the magnitudes of negative entries. The OQ negativity has greater visibility than the MHQ negativity. Both yield exactly the same extractable work because their A and B marginal probabilities coincide. The different negativity magnitudes therefore cannot rank the extracted work.

The joint-measurability criterion and OQ-MHQ equivalence apply to binary measurements in two dimensions. They do not establish the same equivalence for three or more levels. The three-level case needs EPM, TPM, and weak-TPM to reconstruct MHQ. The second-moment bound also depends on a symmetric energy profile and the projective-measurement treatment in the proof.

The nitrogen-vacancy calculation is based on a prior experiment, and this paper reports no new experimental run. The authors cite existing optical and collective-measurement work as evidence that the settings are experimentally accessible. They leave the connection between non-joint measurability and negativity-induced weak values for further study.

For nitrogen-vac-center experiment design, the actionable controls are initial-state coherence, measurement sharpness, and the EPM and TPM sequences. The OQ construction uses EPM and TPM data. Coherent input states require care because sequential TPM removes coherence during the first measurement.

Measurement sharpness μ provides a direct test for the transition from joint to non-joint measurability in binary unbiased qubit measurements. Excess OQ work above max_μWcl certifies non-joint measurability. Quasiprobability negativity magnitude should not be used as a proxy for extractable work.

An NV-center implementation uses diamond as the host for the three-level spin system modeled in this paper. That implementation requires the specified |0⟩, |1⟩, and |−1⟩ states, but the paper omits crystal orientation, nitrogen concentration, defect geometry, sample dimensions, and fabrication method. The diamond materials catalog lists MPCVD-grown single-crystal and polycrystalline diamond, while engineering support can assess compatibility once host and defect requirements are defined.

View Original Abstract

We employ the operational quasiprobability (OQ) as a work distribution, which reproduces the Jarzynski equality and yields the average work consistent with the classical definition. The OQ distribution can be implemented through two simple experimental settings: the end-point measurement and the two-point measurement. Using this framework, we demonstrate the explicit contribution of coherence to the work, which can be formulated by the second-law-type relation. In a two-level system, we show that nonjoint measurability, a generalized notion of measurement incompatibility, can increase the amount of extractable work beyond the classical bound imposed by jointly measurable measurements. We further prove that the real part of Kirkwood-Dirac quasiprobability (KDQ) and the OQ are equivalent in two-level systems, and they are non-negative for binary unbiased measurements if and only if the measurements are jointly measurable. In a three-level nitrogen-vacancy center system, the OQ and the KDQ exhibit different amounts of negativities while enabling the same work extraction, implying that the magnitude of negativity is not a faithful indicator of nonclassical work. These results highlight that coherence and nonjoint measurability play fundamental roles in the enhancement of work.