Sensing Through Relaxation: A Programmable Floquet Response

Most quantum sensors work a little like extremely delicate clocks. A magnetic field or another disturbance changes how quickly a quantum state evolves, and the sensor reads the accumulated change. The difficulty is that environmental noise can scramble the clock before it has gathered enough information.

A new theoretical preprint proposes a different strategy: program the way a driven quantum state relaxes, then infer the disturbance from that relaxation. Instead of asking only how long a spin can maintain its phase, the researchers ask whether the shape of its fading signal can become a useful measurement. The idea turns a familiar limitation into part of the sensing mechanism.

In Programmable Floquet Response for Quantum Sensing, submitted on October 12, 2026, Dionisio Cendoya, Kang Xu, Lisandro Buljubasich and Carlos A. Meriles develop and numerically test this approach. Their affiliations span Argentina’s Universidad Nacional de Córdoba and CONICET, and the City University of New York. This is a theory-and-simulation preprint, not a new experimental sensor demonstration. [1]

The proposal is not to eliminate relaxation. It is to make relaxation respond to a signal in a deliberately programmable way.

From a quantum clock to a quantum response curve

In a conventional Ramsey measurement, a two-level quantum system is prepared in a superposition. Its two components acquire a relative phase during a waiting period. A final control pulse converts that phase into a measurable population difference. The technique is central to quantum sensing, as explained in the established review by Christian Degen, Friedemann Reinhard and Paola Cappellaro. [2]

The available waiting time matters. Noise that changes the system’s transition frequency can randomize its phase. Carefully timed pulses can cancel some disturbances, extending coherence and enabling selective detection of alternating signals. But the same cancellation can also suppress a slowly varying signal that one wants to measure. Sensor design therefore involves choosing which changes to reject and which to retain.

The October proposal changes the observable. Rather than preserving a phase throughout the measurement, it uses periodic control to create a driven state with a particular preferred orientation. A signal shifts the system’s resonance frequency, changes that orientation and, through a repeated transfer protocol, alters how quickly the measured polarization declines.

What Floquet engineering contributes

Floquet theory describes quantum systems whose governing conditions repeat in time. Here the repetition is a block of control pulses. Within a block, pulses may differ in phase, amplitude, frequency and spacing. Repeating the block produces an effective description of the evolution observed once per cycle.

For the spin considered in the paper, that description includes an effective Floquet locking field. This is a mathematical representation of the driven spin’s preferred axis and evolution, not an additional permanent magnet placed beside the sample. The pulse sequence determines its properties. Crucially, both its direction and its strength depend on the frequency offset between the spin resonance and the applied control. [1]

That offset is called detuning. If a magnetic field shifts the spin’s resonance while the control settings remain fixed, the detuning changes. The effective locking direction changes with it. The researchers use this dependence to translate a small external perturbation into a measurable response.

Conventional spin locking already uses sustained driving to establish a preferred axis in a rotating reference frame. The new proposal builds on that background rather than inventing driven-spin sensing from scratch. Its emphasis is the additional freedom provided by structured, periodically repeated pulse blocks and transfers between their distinct locking conditions.

Two locking conditions make small mismatches accumulate

The most useful way to picture the protocol is as a repeated handoff. The sensor first settles under pulse block A. A control operation then transfers its polarization into the locking basis associated with block B. At a chosen reference resonance, that handoff is designed to align correctly.

When the resonance shifts, the two locking directions change. A transfer calibrated for the original resonance is no longer perfectly matched. Part of the transferred polarization points across the new locking axis rather than along it. During the settling stage, that transverse component relaxes, reducing the remaining locked polarization.

Repeat the transfer-and-settle process, and small mismatches accumulate into a larger population contrast. The model therefore converts a resonance shift into a change in a relaxation trajectory. In the examples studied, the researchers simulate environmental fluctuations with a Lorentzian noise spectrum and average over stochastic realizations. They are calculating a response under a specified noise model, not reporting measurements from a finished instrument. [1]

Why the sensor should not sit exactly at perfect alignment

Perfect transfer sounds ideal, but it is not necessarily the most sensitive operating point. Near exact alignment, the transfer loss can change only quadratically with a small detuning shift. The response is then relatively flat at its centre, making tiny signed changes harder to distinguish.

The paper instead introduces a deliberate bias away from the reference point. Around this working point, small resonance changes produce an approximately linear signal change. Choosing it involves a trade-off: stronger mismatch dependence improves responsiveness, but excessive loss shortens the useful measurement interval.

The number of useful handoffs is not unlimited. The driven locked state has a finite lifetime, commonly described by a rotating-frame relaxation time. Pulse errors, environmental coupling and changing detuning can affect that lifetime. This is a route to using different dynamical timescales, not a claim of an indefinitely stable sensor.

Reading the fading curve rather than its brightness

A simple implementation would measure population after a selected interrogation time. A more interesting option is to measure several points along the relaxation curve and estimate its characteristic decay rate. That distinction is central to the paper’s practical argument.

Imagine a light signal that fades with time. If the detector gain changes between runs, a measurement of brightness at one instant may confuse that gain change with the quantity being sensed. A fit to the whole curve can, under suitable conditions, separate its overall amplitude from the timescale of the fading.

The analogy is imperfect but useful. In the proposed sensor, changes in optical readout contrast, amplitude or baseline can distort a population signal. If those disturbances remain approximately constant during each individual relaxation trace, fitting the temporal response can reduce their influence on the inferred relaxation rate. Fast disturbances within a trace, or changes that alter the dynamics themselves, need separate treatment.

A more stable long-term estimate can be worth a worse short-term sensitivity—but only when the relevant instrumental drift is actually present.

The authors explicitly report this trade-off. Under ideal measurement conditions, estimating a resonance shift from a relaxation trajectory generally carries a statistical penalty compared with conventional Ramsey sensing. In their simulations with slow instrumental fluctuations, Ramsey sensing initially performs better, but eventually becomes limited by drift. The relaxation-based estimate continues improving and can overtake it at longer averaging times. [1]

This does not establish universal superiority over Ramsey methods, nor over every drift-corrected experimental implementation. It establishes a reason to compare methods using realistic instrument noise and total measurement time, rather than ranking them only by an idealized short-time sensitivity.

From steady fields to selected alternating signals

The same framework can also respond selectively to alternating perturbations. The researchers organize transfers into larger cycles with opposite bias detunings. These symmetry-related sequences respond oppositely to a resonance shift. Timing their alternation to an external oscillation creates frequency selectivity.

In familiar pulse-based magnetometry, sequence timing likewise determines which frequencies are detected. Here the selectivity is built into a larger Floquet transfer cycle while preserving the cumulative relaxation mechanism. The paper identifies comparatively low frequencies, typically below roughly 50–100 kilohertz, as a promising operating region for the proposed design. That is a proposed regime, not a measured instrument bandwidth or a universal boundary for competing sensors. [1]

A narrow frequency response can help separate a target signal from neighbouring disturbances. It also demands attention to drift in the signal frequency and the time needed to gather data. A sharper filter is useful only when its measurement cost and stability fit the actual application.

Why a quantum-energy research hub should care

This study is about sensing, not energy generation or storage. Its relevance to quantum-energy research is therefore an enabling connection, and the following possibilities are editorial analysis rather than demonstrated applications.

Diagnosing driven devices

A quantum heat engine, battery or driven material needs reliable characterization. Magnetic-field drift and changes in local resonance conditions can obscure whether a device has changed or the measurement apparatus has drifted. A sensor that encodes information in a robust temporal response could potentially help distinguish those effects.

That possibility needs experimental testing in the environment of a real device. The sensor’s own drive could disturb the system under investigation, and the target signal may fluctuate too rapidly for the assumed separation of timescales. Better metrology is a plausible connection, not an automatic consequence of the simulation.

Designing useful open-system dynamics

Quantum thermodynamics deals with systems exchanging energy with controls and environments. The proposal illustrates a broader engineering habit: rather than treating every relaxation process as an enemy, design the driven dynamics so that a particular response performs a task.

In this case the task is parameter estimation. In another device it might be state preparation or controlled energy transfer. Those tasks have different performance measures. Improving a sensing estimate does not establish improved heat-to-work conversion, and observing relaxation does not by itself measure a heat current.

What would make the next result convincing?

The immediate milestone is an experimental demonstration of the full response-engineering protocol. A useful test would measure the locking-axis dependence, show cumulative transfer mismatch and compare rate-based estimates with a carefully optimized conventional baseline under controlled drift.

• Matched resources: compare methods using the same overall observation time, preparation overhead and readout opportunities.
• Measured noise: characterize which fluctuations are slow relative to a trace and which change the relaxation rate itself.
• Control robustness: test pulse errors, drive-frequency instability and changes in environmental coupling.
• Honest bandwidth: report response width, usable frequency range and the time required to resolve a signal.
• Energy accounting: if efficiency becomes part of the claim, measure control and readout consumption alongside achieved precision.

There is also room for optimization. The authors suggest that machine-learning-assisted searches could explore pulse sequences tailored to a signal and its noise environment. That is a proposed development direction, not evidence that an AI-designed sensor has already achieved an experimental advantage.

The central contribution is a shift in the design question. Instead of asking only how to keep a quantum clock coherent for longer, one can ask how to program a quantum system’s entire response to a disturbance. If experiments validate the proposed trade-off, Floquet engineering could provide a useful way to build sensors that remain informative when slow instrumental drift makes the most sensitive short-term method unreliable.