A signal hidden in the photon
The study asks where information about a magnetic field is stored after a single photon interacts with a Zeeman-sensitive three-level system, and whether a detector that resolves the photon's frequency can learn more than one that records only photon loss. The supplied document is a version-1 arXiv preprint dated 28 August 2026. Its analysis follows a modeled scattering process in which the outgoing pulse is described by a magnetic-field-dependent frequency response.
At the start of the calculation, the three-level system is in its ground state, the environment is in vacuum, and the incoming field contains a single-photon pulse. After scattering, the outgoing pulse is described by a frequency-dependent amplitude that changes with the magnetic field. In the model, information can appear in whether the photon remains in the pulse output, in how the output spectrum's intensity is redistributed, and in how its phase changes across frequency.
Counting frequency, not just losses
To compare the possible measurements, the researchers use the output-state quantum Fisher information, or QFI, as the model's full benchmark for estimating the field. They split it into a photon-loss contribution and the QFI of the conditional one-photon output, weighted by the probability that the one-photon outcome occurs. Frequency-resolved photon counting adds something the loss measurement cannot see: its classical Fisher information, or CFI, combines the photon-loss CFI with information from the intensity of the output spectrum. In plain terms, the detector records both whether a photon was lost and how the surviving photon's frequencies are distributed.
That measurement still does not necessarily capture the whole benchmark. The uncaptured part is a nonnegative variance in how the output spectrum's phase responds to the magnetic field across frequency. The gap disappears exactly when that phase response is frequency-independent over the spectrum's support. In the paper's finite-pulse calculation, the gap became small near the reported side maxima, which is why counting could be close to quantum-optimal there.
The tuning that makes counting work
That near match is linked to zeros in the scattering amplitude, meaning frequencies at which the modeled output response falls to zero. The paper finds that such a real-frequency zero is possible at critical coupling, when environmental coupling equals the coupling to the pulse modes, provided a relation between the detunings and control-field strength is met. In the reported calcium-ion parameter example, one zero reaches the pulse center when the reduced field displacement is about 0.574. The calculations therefore tie near-optimal counting to a zero lying within the pulse's bandwidth, not to critical coupling alone.
In one finite-pulse slice, the calculation used critical coupling and set the pulse-duration product to 2. The central point, with zero reduced field displacement, lay between two nearly symmetric information maxima. Near those side maxima, frequency-resolved CFI stayed close to QFI, while photon-loss CFI was smaller. The result is a focused operating-point calculation, not evidence of a universal ranking for every pulse or control setting.
Longer pulses produce a separate scaling result under a specific condition. If the scattering response at the pulse center is smooth and nonzero, the extra spectral-intensity information beyond photon-loss counting and the QFI-to-frequency-counting gap both shrink as the inverse square of pulse duration or faster. As the pulse becomes long, both CFIs approach the QFI. The analysis separately warns that an exact central zero does not fit this smooth, nonzero expansion and needs its own treatment.
Coupling changes where the information lives
The model also challenges a simple assumption about environmental coupling. Total QFI can rise over a finite range as that coupling increases, then fall at large coupling as the scattering response broadens. In the displayed sweep, spectral-intensity information and frequency-resolved CFI peaked when environmental coupling matched pulse-mode coupling. Total QFI peaked at a larger normalized coupling, about 1.34 times the pulse-mode coupling. At zero environmental coupling, the sweep contained phase information only. The setting that is best for frequency-resolved counting, in other words, need not be the setting with the greatest total information.
The detector assumptions are demanding. The ideal frequency-resolved CFI assumes unit detection efficiency and arbitrarily fine spectral resolution. Finite resolution averages over frequency bins, which can reduce the CFI and round off the sharp cusp in the ideal calculation. The near match is therefore conditional on resolving the spectral structure that carries the extra information.
Taken together, the preprint describes a conditional advantage for measuring a photon's frequency. That measurement can add spectral-intensity information beyond photon-loss detection, and the reported finite-pulse calculation puts it close to the full QFI near selected side maxima. But the result depends on the placement of scattering zeros, pulse duration and detector resolution. The supplied document is an arXiv version-1 preprint dated 28 August 2026, so the evidence presented here is the model's analytical and numerical result, not a measured performance claim.
Paper data and sources
Original title: Pulsed single-photon magnetometry with a $Λ$-type three-level system: near-optimal frequency-resolved photon counting
Authors: Seyed Mostafa Moniri, Elnaz Darsheshdar, Mikayel Khanbekyan
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-28
DOI: Not available
Original paper · Full text