Why in situ characterization matters for photon-subtraction devices
Photon subtraction is a workhorse for generating non-Gaussian optical states, and its performance hinges on two parameters: the beam splitter transmissivity τ and the heralding detector's quantum efficiency η [1]. Classical calibration with bright beams fails to capture low-photon-flux behavior, and manufacturer values can drift with wavelength, power, temperature, and spatial mode [1]. Full quantum state tomography could extract these parameters but typically requires a large measurement set and offers limited precision, making it impractical for routine characterization [1]. Earlier work established that click statistics from on/off detectors can genuinely characterize light states without relying on high detection efficiency [5], and precursor methods showed that quantum state reconstruction can proceed even with imperfect detection [3]. The new paper extends this lineage by proposing an in situ scheme that uses only the heralding clicks and homodyne data already generated during standard operation, eliminating extra calibration overhead [1].
How the protocol decouples transmissivity and detector efficiency
The scheme feeds a displaced squeezed state into the beam splitter, performs homodyne detection on the transmitted mode, and records on/off clicks on the reflected mode [1]. The click statistics constrain the product η×τ, while the conditional homodyne distribution of the transmitted mode reveals the conditional statistics needed to separate the two parameters [1]. The authors derive the total Fisher information matrix by combining the information from the detector click statistics with that from homodyne measurements conditioned on both 'off' and 'on' events [1]. For 'off' events the conditional state remains Gaussian, yielding an analytic Fisher information; for 'on' events the state is non-Gaussian, so the Fisher information is evaluated numerically from the definition [1]. This dual-channel approach is what enables simultaneous estimation without dismantling the device [1].
Sloppiness analysis identifies the optimal homodyne quadrature
A central quantity in the analysis is the sloppiness parameter S = 1/det(F), which quantifies parameter degeneracy: large S signals a nearly singular Fisher information matrix and poor simultaneous identifiability, while small S indicates both parameters can be estimated well [1]. The authors find that sloppiness always attains its global minimum at homodyne phase φ = 0, corresponding to amplitude quadrature measurement, across different combinations of τ, η, probe photon number N, and squeezing fraction β [1]. This minimum persists for both low-photon (N = 1) and high-photon (N = 10) probes, and for coherent-state probes (β ≈ 0) as well as highly squeezed probes (β = 0.9) [1]. The practical implication is that choosing the amplitude quadrature substantially reduces parameter degeneracy and enables efficient simultaneous estimation [1].
Joint estimation outperforms sequential estimation over a broad parameter range
The paper compares joint estimation of τ and η with a sequential strategy in which one parameter is estimated first and then the other [1]. Using optimized operating parameters, the joint estimation strategy consistently yields a lower estimation bound than the sequential approach over the investigated parameter range [1]. The total variance bound Tr[Cov(θ)] ≥ Tr[F⁻¹]/M is used as the figure of merit, where a smaller value indicates better overall precision [1]. The authors also analyze how probe photon number, squeezing fraction, beam splitter transmissivity, and detector efficiency affect the achievable bounds, identifying operating regimes that maximize precision [1]. This comparison matters because sequential calibration is the intuitive default, and the result shows that simultaneous estimation is more efficient for this device [1].
Boundaries of the conclusion and what remains uncertain
The conclusions rest on theoretical Fisher information and numerical analysis, not on experimental implementation [1]. The protocol assumes a displaced squeezed state with real squeezing parameter and specific detection conditions; its performance outside these assumptions is not established [1]. The analysis also assumes classical multi-parameter estimation theory and does not aim for ultimate quantum-limited precision or novel Cramér–Rao bounds [1]. Dark counts, detector dead time, and other experimental imperfections are not modeled in the supplied analysis, though earlier work on click detectors has shown that dark counts can be neglected in some pulsed regimes but add with multiplexed detectors [4]. Validation evidence from multiplexed on/off detection demonstrates that click statistics can reliably characterize non-classicality without corrections or post-processing [5], but whether the same robustness carries over to this specific joint estimation scheme remains an open experimental question [1].
About These Sources
This research page is built on 5 studies (4 peer-reviewed, 1 preprint) — published from 2014 to 2026, 2 from 2024 or later, collectively cited 60 times — selected as the most relevant from 7 studies that passed quality screening, drawn from 62 papers retrieved from a database of over 500 million.
Sources used in this answer
In situ characterization of a photon-subtraction device via heralding counts and homodyne detection
The anchor paper proposes an in situ scheme for simultaneous estimation of beam splitter transmissivity and heralding detector quantum efficiency using click statistics and homodyne measurements, deriving the Fisher information matrix and showing joint estimation outperforms sequential estimation over a broad parameter range.
A comprehensive review of photon subtraction: principles, implementations, and emerging applications in quantum technologies
A comprehensive review establishes photon subtraction as a technique that projects a transmitted field into a non-Gaussian state via a low-reflectivity beam splitter, providing foundational context for the device architecture.
Tomography by Noise
A precursor work on tomography by noise shows that traditional methods require high detection efficiency and precise calibration, motivating alternative characterization approaches that tolerate imperfect detection.
Quantum illumination with multiplexed photodetection
Validation evidence on quantum illumination with multiplexed photodetection shows that on/off click detectors are useful for heralding and that dark counts can be neglected in some pulsed regimes but add with multiplexed detectors, informing practical limitations.
Harnessing click detectors for the genuine characterization of light states
Validation evidence on harnessing click detectors for genuine light-state characterization experimentally demonstrates that multiplexed on/off detector click statistics can reliably identify non-classicality without corrections or post-processing, supporting the generalizability of click-based methods.