When Quantum Engines Fluctuate: What Does Efficiency Really Measure?

A tiny heat engine does not necessarily deliver the same work every time it runs. At quantum scales, even an accurately repeated control sequence can produce a distribution of outcomes. An October 2026 preprint asks how moving a quantum piston faster changes that distribution—and why a single headline efficiency can conceal the behaviour that matters most.^[1][2]

Bikram Keshari Behera and Shyamal Biswas study a single particle confined inside a one-dimensional box with a moving wall. They calculate work and efficiency fluctuations for Carnot-type and Otto cycles, combining finite-speed isolated strokes with idealized thermal-contact stages. This is an analytical and numerical study, not a new experimental engine or a demonstrated power source.^[2]

The useful question is not simply “How efficient is the engine?” It is “Which efficiency is being measured, how variable is its output, and what resources are counted?”

The distinction is especially important for beyond-Carnot discussions. A large value in a microscopic efficiency distribution is not, by itself, evidence that a machine beats the thermodynamic limit in sustained operation. Earlier peer-reviewed research already shows why fluctuating ratios must be interpreted separately from the ratio of average work to average heat.^[3]

A quantum piston without a miniature car engine

The model’s “piston” is a moving boundary that changes the space available to a quantum particle. Instead of following a sharply defined classical trajectory, the particle occupies quantum energy states. Changing the box length changes their energies; moving the wall at finite speed can also produce transitions between those states. The preprint calculates those transition probabilities and uses them to construct work statistics.^[2]

Think of changing the dimensions of a musical instrument. The available notes change when its geometry changes. A quantum engine adds another question: after the geometry is altered, which of the new energy states will be occupied? This analogy is only a guide to the changing spectrum, not a claim that the particle behaves like an ordinary vibrating string.

A small model can be valuable precisely because it strips away complications. Here, the moving boundary supplies a concrete control parameter—wall speed—while the particle’s discrete energy levels make the possible measurement outcomes explicit. The authors can therefore connect a change in driving speed with a change in their calculated efficiency distribution.^[2]

Two cycles, one finite-speed question

The Otto version

A quantum Otto cycle alternates isolated changes of the working medium’s Hamiltonian with thermal-contact stages at fixed Hamiltonian. In this piston model, the isolated stages change the box length, while the fixed-length stages exchange heat with the hot or cold reservoir. This separates the intended work and heat strokes relatively cleanly.^[2][4]

The Carnot-type version

The Carnot-type construction instead uses quasi-static isothermal stages: the system remains in thermal equilibrium with a reservoir while its geometry changes. These stages are linked by isolated expansion and compression at finite wall speed. The paper calls this a quantum Carnot engine, but the finite-speed protocol should not be confused with a fully reversible Carnot cycle.^[2][4]

There is an important terminology trap. In thermodynamics, “adiabatic” means no heat exchange. In quantum dynamics, adiabatic evolution usually means that populations follow instantaneous energy levels without transitions. An isolated stroke can satisfy the first meaning while failing the second. The new study’s non-quasi-static adiabatic strokes are isolated, yet their finite-speed motion generates quantum transitions.^[2][4]

How the researchers assign work

The preprint uses a two-point energy-measurement scheme on the isolated strokes. Measure energy before moving the wall, evolve the particle under the specified driving, then measure energy again. The difference between the measured values determines the work associated with that realization, with the sign chosen according to whether work is supplied or extracted.^[2]

Two sources of variation enter this description. Thermal preparation determines how likely the particle is to start in each energy state. The subsequent quantum dynamics determines the probabilities for transitions to final states. Repeating the same protocol therefore produces a statistical collection of energy changes, rather than necessarily reproducing one work value.^[2][3]

The measurement prescription is part of the definition, not a detail to hide. A projective energy measurement can change a quantum state, particularly when that state contains coherence between different energies. Reviews of quantum thermodynamics explicitly discuss this limitation. Results obtained with energy measurements should not automatically be identified with the behaviour of an unmeasured coherent device.^[4]

For readers assessing a future experiment, that suggests a practical question: does the reported quantity come from direct repeated energy measurements, an indirect reconstruction, or a theoretical distribution? These can all be useful, but the experiment must explain how its measurement procedure connects to its performance claim.

The definition that changes the story

The most consequential detail in the October preprint is easy to miss. For its Carnot-type engine, Equation 13 divides fluctuating work by the average heat input. The authors justify averaging the thermal-contact contributions through their quasi-static treatment. This produces a work-fluctuation statistic normalized by a fixed heat scale.^[2] Replacing heat by its mean is the authors’ modeling choice; quasi-static thermal contact does not generally eliminate trajectory-level heat fluctuations.^[3][4]

That is not the same object as dividing each cycle’s work by that same cycle’s fluctuating absorbed heat. Denzler and Lutz’s 2020 paper, Efficiency fluctuations of a quantum heat engine, explicitly treats a stochastic work-to-heat ratio and distinguishes it from thermodynamic efficiency, which uses averaged work and heat.^[3]

• Thermodynamic efficiency: average work output divided by average heat input, for the stated operating regime and resource accounting.
• A fluctuating work-to-heat ratio: work and heat from individual realizations, requiring their joint statistics and careful treatment of zero heat.
• Work normalized by mean heat: fluctuating work divided by a fixed average heat input, as used in the preprint’s Carnot-type efficiency definition.^[2][3]

These distinctions are not semantic. Denzler and Lutz show that nonadiabatic driving can produce events with zero absorbed heat and nonzero work in their Otto-engine statistics. The corresponding stochastic ratio is undefined, and its mean can diverge. A definition with a fixed nonzero mean-heat denominator does not have that same event-by-event singularity.^[3]

Consequently, two publications can both discuss “efficiency fluctuations” while plotting different observables. Comparing the heights, tails or averages of their distributions without checking the denominator would be misleading. For this new preprint, the defensible reading is to keep its own definition visible rather than importing conclusions from a differently defined efficiency distribution.

What the October study actually adds

The new contribution is the finite-speed, moving-boundary setting and its comparison of the two cycle constructions. Behera and Biswas report that piston speed introduces additional quantum fluctuations that modify the calculated average efficiencies. Their distributions contain discrete peaks linked to allowed energy differences and transition probabilities.^[1][2]

The paper also explores temperature dependence and the approach to quasi-static and classical limits. Its conclusion states that familiar equilibrium results are recovered in the appropriate combined limits. This is a model-dependent analysis of how a quantum working medium responds to driving, not a universal numerical efficiency gain available to every quantum engine.^[2]

Those reported trends should remain attributed to a preprint. This article does not independently reproduce its numerical distributions or establish their robustness to different measurement, thermalization or control models. In particular, the probability weighting and thermalization assumptions need independent checking. Equation 12 weights both endpoints of each isolated transition with Gibbs probabilities, unlike the conditional-probability construction used in the established framework; its normalization therefore deserves scrutiny.^[2][3] Changing the treatment of heat fluctuations would also change the statistical question being answered.

The authors themselves identify extensions still to be studied: many-particle working media and non-quasi-static thermal-contact stages, among others. They note that finite-time treatment of those thermal stages would introduce heat-input fluctuations and lead to a more strongly nonequilibrium problem.^[2]

Why this is not a Carnot violation

For a cyclic heat engine powered only by two equilibrium thermal reservoirs, the usual Carnot bound constrains thermodynamic efficiency. Quantum mechanics does not remove that accounting requirement. The broad review by Cangemi, Bhadra and Levy distinguishes conventional thermal operation from machines supplied with extra resources such as nonthermal reservoirs, coherence or measurement-and-feedback processes.^[4]

A rare or unusually large microscopic ratio is a different kind of statement. It describes a selected outcome or statistical variable, not automatically the net conversion achieved over repeated operation. Likewise, calling a protocol “Carnot” does not establish that every finite-time version is reversible or saturates the Carnot bound.^[3][4]

This is not an argument against quantum advantages. It is a way to distinguish a genuine improvement against a specified classical or quantum benchmark from a change of definition. The review literature documents several distinct meanings of quantum advantage; choosing the relevant comparison is part of the scientific task.^[4]

The connection to Floquet engineering

This particular preprint is not presented as a Floquet-materials experiment. Its connection to Floquet engineering is a shared control problem: externally changing a quantum Hamiltonian affects energy-state transitions, work and fluctuations. A repeated engine protocol also requires care about how its isolated dynamics and reservoir contacts fit together.^[2][4]

Our interpretation is that the piston model offers a useful diagnostic mindset for periodically driven energy research. Rather than optimizing only a mean value, researchers can ask how a drive redistributes the possible outcomes. Does it create a more predictable output, increase unwanted transitions, or improve one metric while worsening another?

A full Floquet analysis would require its own specified periodic evolution and, where relevant, a treatment of dissipation and the long-time operating state. The present study should not be relabelled as such a calculation. Its value for this audience lies in the questions it makes unavoidable, not in an unreported Floquet result.

What a useful next test should report

For a follow-up simulation or experiment inspired by this work, we would prioritize four checks. These are editorial recommendations, not results already demonstrated by the preprint:

• Repeatability: show the work distribution alongside the average output, rather than publishing only the best-looking efficiency number.
• Consistent definitions: state whether heat fluctuates and whether efficiency means a ratio of averages or a statistic built from individual realizations.
• Complete timing: include thermal-contact and reset times before drawing conclusions about power. Speeding one stroke is not equivalent to increasing full-cycle power.
• Complete resource costs: identify the control, measurement and preparation resources required for the reported operating mode.

The practical lesson is modest but important. Before quantum engines can be judged as energy technologies, their performance must be described in terms that survive changes in measurement protocol and accounting. The October piston preprint provides a fresh theoretical setting for that discussion. Its strongest takeaway is not “more than Carnot,” but a sharper question: what exactly does an efficiency distribution tell us about usable, repeatable work?