The quantum harmonic Otto cycle

I. INTRODUCTION

• The review focuses on the quantum harmonic oscillator as a working medium, a model that remains analytically tractable while representing generic quantum-engine phenomena.The introduction frames the review as an analysis of recent progress on this working medium.Source: §I · pp. 4–5
• A single-ion heat engine with an effective harmonic-trap frequency has been experimentally realized and can be classified roughly as a reciprocating Otto engine.Its operation periodically modulates the trap frequency.Source: §I · p. 4
• Quantum reciprocating engines using a qubit showed performance resembling macroscopic engines, with endoreversible efficiency at maximum power converging at high temperature to the Novikov–Curzon–Ahlborn prediction.Deviations were reported as small even at low temperature, despite a different heat-transport law.Source: §I · pp. 4–5
• The oscillator provides enough complexity to represent quantum analogues of friction and heat leaks while remaining analytically manageable.The introduction links such phenomena to a control Hamiltonian that does not commute with the internal Hamiltonian, producing non-adiabatic effects.Source: §I · p. 5
• Under weak system–bath coupling, quantum open-systems theory establishes consistency between quantum mechanics and thermodynamic laws, allowing work and heat to be defined in the quantum regime; at strong coupling, this connection remains unestablished.Source: §I · p. 5

II. THE QUANTUM OTTO CYCLE

• The quantum Otto cycle uses hot and cold isochores for heat exchange and isolated expansion and compression adiabats for work, with each stroke represented by a propagator.The oscillator frequency changes from ωh to ωc during expansion and returns to ωh during compression.Source: §II · pp. 6–7
• The cycle map is a composition of completely positive stroke propagators, whose order matters because the segment propagators do not commute.The same operators with different parameters can also describe an Otto refrigeration cycle.Source: §II · p. 7
• In the adiabatic limit, the population N stays constant during expansion and compression, and the cycle’s work and efficiency are then specified under that condition.The supplied passage identifies Δω = ωh − ωc and a periodic-cycle population difference ΔN = ΔNc = ΔNh, but the expressions themselves are not included.Source: §II · p. 7

A. Quantum Dynamics of the Working Medium

• A harmonic oscillator serves as an analytically tractable working medium, with its trap parameter controlled to drive the cycle.Its Hamiltonian is generated by position and momentum operators; the frequency changes between hot and cold values on the adiabats.Source: A. Quantum Dynamics of the Working… · pp. 7–8
• Time-dependent, noncommuting Hamiltonians govern unitary adiabats, while finite-time hot and cold isochores use Lindblad dissipation toward thermal equilibrium.The noncommutation of kinetic and varying potential energy is identified as the origin of quantum friction [39].Source: A. Quantum Dynamics of the Working… · p. 9

III. QUANTUM THERMODYNAMICS

• The Otto-cycle energy balance assigns power production or consumption to the adiabats and heat transfer to the isochores.The first law is expressed as dE = dW + dQ, with energy given by the expectation of the Hamiltonian.Source: §III · pp. 10–11
• Canonical invariance preserves a generalized canonical state when its observables are closed under the dynamics; a closed Lie algebra provides this closure on unitary adiabats.The resulting thermodynamical observables determine the state under the maximum entropy principle.Source: §III · pp. 11–12
• For the harmonic working medium with LD defined in (9), canonical invariance of the state holds throughout the cycle [24].The condition is that LDρρ−1 is a linear combination of operators in the algebra.Source: §III · p. 12
• Representing the state and cycle propagators on a thermodynamic-observable vector space reduces the propagator dimension from N^2 to M, where N is Hilbert-space size.The operator-space dimension M is determined by the chosen thermodynamically significant observables.Source: §III · p. 12
• For the oscillator, the selected thermodynamic variables are the Hamiltonian H, Lagrangian L, and position-momentum correlation C = ωD.These variables are linear combinations of the same Lie algebra as Q^2, P^2, and D.Source: §III · p. 13
• The Casimir operator commutes with all operators and remains constant during unitary segments generated by H; a companion and further invariant are also defined.The harmonic-oscillator Casimir is described as a positive operator.Source: §III · p. 13
• Increasing coherence has an energy cost, and coherence is characterized by deviation from a state diagonal in energy [61].The supplied passage does not give the full expression for the energy change.Source: §III · p. 13
• The canonical state is parameterized by β, γ, and γ*, and is equivalent to a squeezed generalized Gibbs state with non-commuting constraints [65].Examples of these states have Gaussian phase-space shapes.Source: §III · pp. 13, 15
• On the hot isochore, the system exhausts heat and equilibrates, spiralling from high energy toward zero correlation and Lagrangian [Figure 2].The cycle description also shows energy loss on the expansion adiabat, energy gain from the cold bath, and return along the compression adiabat.Source: §III · p. 14, Figure 2 · p. 14

• The von Neumann entropy, S_VN = −tr{ρ ln ρ}, depends only on the state and is the minimum information needed to specify ρ, with S_VN ≤ S_O.Source: Entropy Balance · p. 15
• Energy-measurement entropy S_E is at least S_VN and equals it only when ρ is diagonal in the energy basis, as in thermal equilibrium.Source: Entropy Balance · p. 16
• The relative entropy between ρ and its energy-diagonal state ρ_ed measures coherence and equals S_E − S_VN ≥ 0.Source: Entropy Balance · p. 16
• Von Neumann entropy is invariant under unitary evolution because unitary transformations preserve the eigenvalues of ρ.Source: Entropy Balance · p. 16
• For a nonequilibrium oscillator, S_E equals the entropy of an equilibrium oscillator with the same mean energy; a squeezed pure state has S_VN = 0 but S_E ≥ 0.Source: Entropy Balance · p. 17
• The energy-based entropy defines a generalized internal temperature Tint through the oscillator’s energy-temperature relation, and this temperature sets the work required to generate coherence.Source: Entropy Balance · p. 17

IV. THE DYNAMICS OF THE QUANTUM OTTO CYCLE

• A quantum heat engine faces a tradeoff between efficiency and power, and its performance can be optimized for both.Source: §IV · p. 17
• The Otto cycle is partitioned into four strokes that are combined into a full cycle, with each segment affecting power extraction or refrigeration.The cycle can be optimized first by segment and then globally with respect to efficiency and power.Source: §IV · p. 17

A. Heisenberg Dynamics of Thermalisation on the Isochores

• Isochores extract heat from or reject heat to thermal reservoirs while the working medium approaches thermal equilibrium.The Hamiltonian remains constant on each isochore, with ω=ωh/c.Source: A. Heisenberg Dynamics of… · pp. 17–18
• The Heisenberg equations of motion describe the dynamics of an operator X during thermalisation on an isochore.Source: A. Heisenberg Dynamics of… · p. 18
• Isochore evolution yields a propagator on the observable vector space H, L, C, I, with R=e^-Γt and oscillatory factors cos(2ωt) and sin(2ωt).The passages identify Γ as the heat conductance and state that the transition rates obey detailed balance.Source: A. Heisenberg Dynamics of… · p. 18
• The isochore propagator does not generate coherence from energy, and the coherence observables L and C are not coupled to H.The cited passages identify L and C as the observables whose expectations describe coherence.Source: A. Heisenberg Dynamics of… · p. 18

B. The Dynamics on the Adiabats and Quantum Friction

• Noncommuting time-dependent Hamiltonians generate coherence at extra energy cost; this cost is dissipated on the isochores as quantum friction [39].Generating coherence consumes power when the initial state is diagonal in energy.Source: B. The Dynamics on the Adiabats and… · pp. 19, 21
• For |µ| → 0, energy decouples from coherence, frictional power vanishes, and the cycle is characterized by energy-level populations.Away from this limit, the non-adiabatic parameter µ characterizes the dynamical coupling between the operators.Source: B. The Dynamics on the Adiabats and… · p. 19
• At constant µ, the solutions are oscillatory for |µ| < 2, while their trigonometric functions become hyperbolic for |µ| > 2.Source: B. The Dynamics on the Adiabats and… · p. 20
• Frictional energy cost scales as µ^2, so slow operation suppresses it but entails longer cycle times and lower power.Source: B. The Dynamics on the Adiabats and… · p. 21
• Periodic solutions can generate and then consume coherence, making the propagator diagonal when cos(Ωθ(τa)) = 1; the shortest frictionless solution occurs at l = 1.Source: B. The Dynamics on the Adiabats and… · p. 21
• Finite-time frictionless controls, or shortcuts to adiabaticity [79] [81] [82], return the state diagonal in energy at the isochore interface, allowing the engine to be described by energy populations.For the harmonic working medium, the number-operator expectation value also characterizes the engine.Source: B. The Dynamics on the Adiabats and… · pp. 21, 23
• An invariant-operator approach uses Ermakov-equation solutions to construct fast frictionless strokes, subject to constraints such as real, positive frequency.The minimum time depends on the imposed constraints; inverted harmonic potentials with imaginary frequency permit faster strokes.Source: B. The Dynamics on the Adiabats and… · p. 22
• Optimal control minimizes time with a control linear in ω(t); bounded frequencies yield bang-bang switching, while other constraints can make energetic cost diverge.With ωmax = ωh and ωmin = ωc, the optimal time has the stated frequency-ratio scaling; reasonable control constraints give τ* = O(1/√(ωcωh)).Source: B. The Dynamics on the Adiabats and… · pp. 22–23

Amplitude noise

• Amplitude-noise friction diverges as l → ∞ or µ → 0, even for very small γa; the shortest frictionless protocol best suppresses this noise.For large l, F = (ωh − ωc).Source: C. The Influence of Noise on the… · p. 24

Phase noise

• At first order in µ, phase noise preserves δf(1) = 0, with the asymptotic correction δf(1) = eγpFµ −1 ≈ 0.Phase noise arises from errors in the durations of time intervals used to schedule ω(t), modeled as Gaussian white noise equivalent to dephasing on the adiabats.Source: C. The Influence of Noise on the… · pp. 24–25
• Second-order phase-noise evolution mixes energy and coherence even as µ → 0 and τa → ∞, with δf(2) = cosh(β) − 1 > 0.These limits would otherwise be expected to yield frictionless solutions.Source: C. The Influence of Noise on the… · p. 25

Noise effects

• Imperfect adiabat control always gives δf > 0 and requires additional work against friction.The amplitude-noise model uses a frictionless protocol with constant µ and treats frequency-control fluctuations as Markovian Gaussian white noise.Source: C. The Influence of Noise on the… · pp. 23, 25

D. The Sudden Limit

• When adiabat time vanishes, the propagator reaches the sudden-quench limit, and a compression ratio different from 1 mixes H and L to generate coherence.The sudden propagator is also part of frictionless bang-bang solutions.Source: D. The Sudden Limit · p. 26

E. Effects of an Exceptional Point on the Dynamics on the Adiabat

• For µ ≤ 2, the adiabat generator produces a unitary propagator, while its eigenvalues become real for µ ≥ 2.The transition is possible because the generator is non-Hermitian.Source: E. Effects of an Exceptional Point on… · p. 27
• At µ = 2, the adiabat propagator’s eigenvalues become degenerate; beyond the exceptional point, its dynamics changes from oscillatory to exponential.At the exceptional point, the non-Hermitian generator has a single self-orthogonal eigenvector. The authors also expect exceptional points in the cycle propagator to indicate a drastic change in cycle performance.Source: E. Effects of an Exceptional Point on… · p. 27
• Rescaling time reveals an exceptional point at |µ| = 2 in the classical parametric oscillator, marking the transition from underdamped to overdamped motion.The original model has no dissipation, although the rescaled equation takes the form of a damped harmonic oscillator.Source: E. Effects of an Exceptional Point on… · p. 27

V. CLOSING THE CYCLE

• Combining the four stroke propagators closes the cycle as either an engine or a refrigerator, depending on the parameters, with ωh > ωc in both cases.Frictionless cycles operate as either engines or refrigerators.Source: §V · pp. 27–28
• If internal friction dominates, either cycle can instead dissipate power to both baths.This dissipative engine mode occurs when the oscillator’s internal temperature after the expansion adiabat exceeds Th; the refrigerator mode occurs when it exceeds Tc.Source: §V · p. 28

A. Limit Cycle

• A limit cycle is a periodic steady state with no net energy or entropy accumulation; shortening the cycle adds frictional heat, widening bath temperature gaps and increasing dissipation.Overdriving can dissipate heat to both baths while consuming power.Source: A. Limit Cycle · p. 28

limit cycle.

• If the cycle’s completely positive map has a single invariant state, any initial state converges to it; studied reciprocating engines had one non-degenerate eigenvalue of 1.The invariant state is the limit cycle fixed point of the cycle map.Source: limit cycle · p. 29

B. Engine Operation and Performance

• For adiabatic and frictionless cycles, performance is completely determined by the energy at the switching point between strokes.Source: B. Engine Operation and Performance · p. 30

1. Optimizing the Work per Cycle

• Infinite-time adiabatic operation maximizes work and reaches Otto efficiency equal to Carnot efficiency, but its infinite cycle time makes power zero.No coherence is generated in this limit, so the cycle is described by changes in energy.Source: §1 · p. 30

2. Optimizing the Performance of the Engine for Frictionless Conditions

• Frictionless solutions permit finite-time cycles with the same Otto efficiency as adiabatic operation, using coherence to cancel friction.The additional average energy invested during the adiabat to generate coherence is described as a catalyst because it is not dissipated.Source: §2 · p. 30
• Power optimization requires a finite cycle time and varies the time allocations across engine segments; in the frictionless regime, thermalization on the isochores produces entropy.The resulting optimization is reported to be close to the unconstrained optimum, especially at low temperatures.Source: §2 · p. 31
• In the frictionless limit cycle, the working-medium number expectation is unchanged across the hot isochore, and work per cycle depends on the number change between hot and cold isochores.The stated cycle work is Wq = EC − EB + EA − ED = ℏ(ωc − ωh)(N_B − N_D).Source: §2 · pp. 31–32
• At high temperatures, the efficiency at maximum power is the well-known endoreversible-engine efficiency.The results are also said to support the Novikov–Curzon–Ahlbourn result more strongly than its original derivation.Source: §2 · p. 33
• Heat transport is maximized by allocating infinite time to the isochores, with equal hot- and cold-isocore times optimal if and only if Γh = Γc.For small x, the optimal frictionless condition is estimated as τc = τh ≈ 1.Source: §2 · p. 33
• When the heat-transport rate Γ is sufficiently large, optimal power approaches a bang-bang solution with vanishing time allocated to every engine segment.Source: §2 · p. 33
• In frictionless conditions, entropy production follows work’s time-allocation dependence, so maximizing power also maximizes its rate; it remains positive even when excessive compression yields no work.When power is generated, the entropy-production rate is linearly proportional to power.Source: §2 · pp. 33–34
• Frictionless adiabats are obtained using shortcuts to adiabaticity [82], and this shortcut approach extends to working media composed of interacting particles in a harmonic trap [76].A variant adds projective energy measurements before and after each adiabat to measure work output [118]; as a result, the working medium is always diagonal in the energy basis.Source: §2 · p. 34

3. The Engine in the Sudden Limit

• With friction limiting maximum-work operation, efficiency is below the endoreversible efficiency, which is below the ideal Carnot efficiency.The comparison concerns an engine constrained by friction versus one constrained by heat transport.Source: §3 · p. 35
• Work against friction rises with temperature ratio; at the frictionless-optimal compression ratio, sudden work vanishes as friction balances useful work, and beyond it the cycle dissipates.Beyond that limit, the engine generates entropy at both baths.Source: §3 · p. 36
• In the complete sudden limit, all cycle segments are short, coherence survives at all four corners, and exploiting it can partially avoid friction.The allocated times are much shorter than both the oscillator period and the inverse heat-transfer rate.Source: §3 · p. 36
• Even at zero power, such as C = 1, entropy production is positive, reflecting heat leakage from the hot bath to the cold bath.Source: §3 · p. 37
• As hot- and cold-stroke times approach zero, engine power remains finite, reaching the limit of a continuously operating engine.Source: §3 · p. 37
• The extreme sudden engine requires global coherence: the working-medium state is non-diagonal in the energy representation throughout the cycle.Adding dephasing yields no useful power and makes the cycle operate as a dissipator.Source: §3 · p. 37

4. Work Fluctuation in the Engine Cycle

• Work fluctuations can be calculated from energy fluctuations at the cycle’s four corners [120].Source: §4 · p. 37
• For a generalised Gibbs state, the energy variance is related to internal temperature by Var(E) = (kBTint)^2.Source: §4 · p. 37
• With complete thermalization to the bath, the work variance is smallest at the Carnot compression ratio C = Th/Tc.The oscillator reaches the bath temperature in this case.Source: §4 · p. 38
• Generating coherence increases the energy variance.Source: §4 · p. 38

5. Quantum Fuels: Squeezed Thermal Bath

• On the hot isochore, squeezing modifies thermalization and makes the working medium behave as though coupled to a higher-temperature bath, with ⟨ˆH⟩sq ≥⟨ˆH⟩eq.Source: §5 · pp. 38–39
• Transferred coherence increases cycle work when an adiabatic protocol exploits it to reach a lower-energy state diagonal in energy [130].The coherence acts as quantum availability, allowing more work to be extracted from the system.Source: §5 · p. 39

C. Closing the Cycle: The Performance of the Refrigerator

• Refrigeration requires expansion to cool the working medium below the cold bath and compression to heat it above the hot bath.The refrigerator uses the stroke-propagator order in Equation (62); its performance analysis focuses on coefficient of performance and cooling power.Source: C. Closing the Cycle: The Performance… · p. 39
• For ideal cooling, the hot bath should cool the medium near its ground state, and expansion should leave it near its ground-state energy, EC ≈ ℏωc/2.The stated condition for ground-state cooling is ℏωh ≫ kBTh; frictionless expansion solutions can achieve the task in minimum time.Source: C. Closing the Cycle: The Performance… · p. 40

1. Frictionless Refrigerator

• In the infinite-time limit µ → 0, the adiabatic refrigerator reaches Carnot efficiency but its cooling rate vanishes, Rc = 0.Source: §1 · p. 40
• Frictionless points occur when NC = ND (c = 1); the resulting solutions have optimal efficiency with finite power.The passage associates these solutions with µ < 2 and critical frictional points, referencing the cited optimum-efficiency equation.Source: §1 · p. 41
• Superadiabatic protocols and optimal control theory yield frictionless expansion times scaling as τhc ∝ 1/√(ωcωh).The passage also notes that a minimum-time frictionless solution outperforms a linear frequency ramp.Source: §1 · p. 41

2. The Sudden Refrigerator

• Coherence enables finite cooling power as the sudden refrigerator’s cycle time approaches zero, although its sudden-limit cooling rate vanishes at sufficiently low cold-bath temperature.The sudden-limit expression loses meaning below that temperature, so the refrigerator should not be used there.Source: §2 · p. 41

3. The Quest to Reach Absolute Zero

• As Tc approaches zero, cooling power vanishes; high-compression adiabatic expansion takes significant time, while cold-side thermalisation also affects the cooling-rate scaling.At the end of hot thermalisation, the oscillator is very close to its ground state.Source: §3 · p. 42
• The second law requires α ≥ 0, while the first formulation of the third law strengthens this to α > 0, ensuring zero cold-bath entropy production at absolute zero.For α = 0, compliance with the second law depends on entropy production in the other baths compensating for negative cold-bath entropy production.Source: §3 · pp. 42–43
• The unattainability principle rules out cooling to absolute zero in finite time; its rate constraint is ζ = 1 + α − η when cold-bath heat capacity scales as cV ∼ T_c^η.This formulation accounts for cold-bath heat capacity and constrains system–bath interaction and cold-bath properties as Tc approaches zero.Source: §3 · p. 43
• Frictionless solutions yield an upper bound on optimal cooling rate; optimizing Rc with respect to ωc gives ℏωc = kBTc, and α = 1 for frictionless versus α = 2 for optimal-control frictionless scheduling.Source: §3 · pp. 43–44
• If the adiabat is shorter than the minimum shortcut time, the oscillator cannot reach arbitrarily low energy or temperature, and the sudden limit leaves a positive minimum temperature.In this limit, the refrigerator trivially satisfies the unattainability principle.Source: §3 · p. 44

VI. OVERVIEW

• The harmonic Otto cycle is an analytically tractable quantum-device example and has inspired experimental realization.The review notes that ℏ cancels from Heisenberg equations for thermodynamic variables, raising the question of what is quantum in the model and in quantum thermodynamics.Source: §VI · pp. 44–45
• A generalized Gibbs state and operator-algebra approach describe the dynamics using H, L, and C, with the cycle propagator formed by concatenating stroke propagators.The review uses the Heisenberg representation and thermodynamically relevant observables to describe energy and coherence.Source: §VI · p. 45
• Frictionless cycles require coherence to vanish at stroke boundaries, achieved through specially scheduled frequency changes that cancel generated coherence.For frictionless adiabats, optimal thermalization time is finite and thermalization remains incomplete, enabling an estimate of the minimum cycle time; control noise means some friction remains.Source: §VI · pp. 45–46
• Frictionless operation can attain the engine’s maximum efficiency in finite cycle time, whereas reaching Carnot efficiency entails zero power.The review also relates heat-transport entropy production linearly to power and reports that maximum power implies maximum entropy production.Source: §VI · p. 46
• Entropy production is linearly related to power for heat transport, is maximal at maximum power, and remains positive even at zero power.The review associates this zero-power entropy production with the possible cost of maintaining coherence.Source: §VI · p. 46
• Below a minimum adiabat time, the propagator reaches an exceptional point and cycles enter the sudden regime, where producing power requires coherence.Without coherence, the sudden model acts as a dissipator that generates entropy on both baths; in the continuous limit, friction and heat leaks remain unavoidable.Source: §VI · p. 46
• At extremely low temperatures, Otto-refrigerator cooling power is limited by adiabatic expansion or vanishing heat transport as Tc → 0.The review describes this limit as a dynamical version of the Nernst heat law.Source: §VI · p. 47
• The harmonic cycle has served as a template for interacting-particle Otto cycles and multimode systems, with friction-eliminating protocols generalized to many-body dynamics.Source: §VI · p. 47