Memory-assisted squeezed light velocimetry under realistic loss and incoherent noise

Mustafa Gündoğan, Arash Ahmadi, Markus Krutzik · 2026

Read Memory-assisted squeezed light velocimetry under realistic loss and incoherent noise by Mustafa Gündoğan, Arash Ahmadi, Markus Krutzik (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-27723.txt.

Introduction

Mustafa Gundogan, Arash Ahmadi, and Markus Krutzik propose a velocity sensor using squeezed vacuum and a coherent probe in a two-memory Mach-Zehnder interferometer. One memory stays fixed as reference while the other moves during storage, mapping velocity onto a differential phase at readout. Squeezed light beats equal-resource coherent homodyne within a window set by transmission and phase stability, keeping a small advantage at realistic loss.

Catalog record

CREATORMustafa Gündoğan, Arash Ahmadi, Markus Krutzik
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32530
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.27723

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Memory-assisted squeezed light velocimetry under realistic loss and incoherent noise Mustafa Gündoğan, Arash Ahmadi, Markus Krutzik Abstract We propose a velocity sensor based on a two-memory Mach–Zehnder interferometer fed by a coherent probe and squeezed vacuum, read out by balanced homodyne detection. One memory is taken as a stationary reference, while the second memory moves during storage, so that its velocity is mapped onto a differential interferometric phase at readout. The two memories are otherwise assumed identical and are described by a Gaussian write–store–read lifetime together with the associated unconditional noise floor. Using the classical Fisher information, we derive the velocity sensitivity, the transmission threshold required for a target quantum gain, and the optimum storage time. The squeezed scheme improves on equal-resource coherent homodyne within an operating window set mainly by total transmission and phase stability. For representative near-term parameters, unconditional memory noise floors up to about 10 − 1 10^{-1} photons per trial do not by themselves remove the advantage; after optimization the improvement remains at the few-percent level and is limited chiefly by loss. I Introduction Interferometric phase sensing remains one of the most mature settings for quantum enhancement. Injecting squeezed vacuum into the dark port of a Mach-Zehnder interferometer reduces the detected phase noise below the coherent-state limit, but the improvement survives only when attenuation and technical fluctuations remain sufficiently small 8 ; 16 ; 39 ; 37 ; 30 ; 11 ; 38 ; 41 . Optical memories 13 ; 14 ; 33 ; 40 ; 22 ; 31 add another control parameter to this setting. By storing an optical pulse during a controllable period, they allow an external perturbation to accumulate as a phase of the retrieved light, at the price of storage time-dependent loss and memory-generated noise. Related ideas have also started to appear in gravitationally motivated interferometry, where quantum memories act as interrogation elements or time-delay resources in memory-assisted Mach-Zehnder, Hong-Ou-Mandel, and vertically separated clock-like settings 6 ; 18 . A particularly relevant application in this direction is velocity sensing with stored light, which has recently been proposed and experimentally explored in memory-based architectures 2 ; 1 . The basic idea is to map an optical field into a collective excitation, let it evolve during the storage interval, and then read the light out interferometrically. Since such optical memories operates in a phase preserving way, the output light has a fixed phase in the memory frame. However, when physically translated, this corresponds to a phase shift in the lab frame. For a probe of wave number k_{p}=2\pi/\lambda_{p} , with \lambda_{p} the probe wavelength, this velocity-induced phase can be written as \phi_{v}=-k_{p}v\tau, (1) where v is the average velocity during the storage time \tau . Increasing \tau strengthens the transduced phase, but at the same time the retrieval efficiency of the memories decreases. Experiments on phase-preserving optical storage in solids, cold atoms, and warm vapors 45 ; 20 ; 17 ; 24 ; 28 ; 34 ; 35 ; 9 ; 25 , storage of squeezed states 3 ; 23 ; 4 , together with motion-sensitive slow-light and light-dragging measurements 29 ; 10 ; 43 ; 5 and transport of stored light in moving or reconfigured media 27 ; 49 ; 15 ; 32 , provide the experimental background for such a device. Classical laser-Doppler and self-mixing velocimeters remain the relevant nonquantum benchmarks away from the shot-noise limit 36 ; 44 ; 47 . In the present work, we ask whether the memory-assisted velocimetry platform can be pushed beyond the coherent-state shot-noise limit by replacing classical probe light with genuinely quantum input states. More specifically, we consider whether squeezed-input interferometry can retain a metrological advantage once the full sensing cycle is taken into account: propagation through the external optics, storage and retrieval in the memories, and final homodyne readout. This is a stricter question than ideal loss-only interferometry, because in memory-based sensing the same storage period that amplifies the velocity-dependent phase also reduces the retrieved signal and introduces additional incoherent noise associated with the memory operation 19 ; 21 ; 12 ; 46 ; 26 . To this end, here we develop a Gaussian model for a two-memory Mach-Zehnder velocimeter operated near the dark fringe and read out by balanced homodyne detection, with squeezed vacuum injected at the input. The model includes Gaussian memory lifetime, external optical loss, detector inefficiency, interferometer phase noise, squeezing-angle jitter, electronic noise, and the unconditional memory noise floor p_{n} . Both memories are taken to be identical in their efficiency and noise properties. Within this framework, we derive the Cramér-Rao bound for velocity estimation, identify the minimum total transmission required for a prescribed quantum advantage over the coherent-state benchmark, and determine the optimum storage time when the memory efficiency follows a Gaussian decay law. II Model II.1 Interferometer configuration and velocity encoding The sensing geometry is shown in Fig. 1 . A bright coherent state |\alpha\rangle with N\equiv|\alpha|^{2} photons per interrogation enters input mode \hat{a}_{0} . A single-mode squeezed vacuum state S(r)|0\rangle enters the unused input mode \hat{b}_{0} , where S(r)=\exp\!\left[\frac{r}{2}\left(\hat{b}_{0}^{2}-\hat{b}_{0}^{\dagger 2}\right)\right] (2) is the squeeze operator and r\geq 0 is the squeezing parameter. We use quadratures \hat{X}=\hat{b}+\hat{b}^{\dagger} and \hat{P}={(\hat{b}-\hat{b}^{\dagger})}/{i} for which vacuum has unit variance, \mathrm{Var}(\hat{X})=\mathrm{Var}(\hat{P})=1 . The injected squeezed state is aligned so that the phase-relevant quadrature is squeezed at the source, \mathrm{Var}(\hat{P}_{b_{0}})=\exp(-2r) . The first 50{:}50 beam splitter produces arm modes \hat{u}_{0}=\frac{\hat{a}_{0}+\hat{b}_{0}}{\sqrt{2}},\qquad\hat{\ell}_{0}=\frac{\hat{a}_{0}-\hat{b}_{0}}{\sqrt{2}}, (3) where \hat{u}_{0} is the upper arm and \hat{\ell}_{0} is the lower arm. Both arms are written to and read from identical optical memories. We denote the upper and lower memories by \mathrm{QM}_{U} and \mathrm{QM}_{L} , respectively. Experimentally, each interrogation would consist of writing \hat{u}_{0} into \mathrm{QM}_{U} and \hat{\ell}_{0} into \mathrm{QM}_{L} , storing both modes for the same time \tau , translating only \mathrm{QM}_{U} with average velocity v during storage, and then retrieving and recombining the two fields. The motion changes only the accumulated phase; the memory efficiencies and noise floors are otherwise taken identical. Let \phi_{u} and \phi_{\ell} denote the total phases accumulated by the upper and lower arms between the first and second beam splitters. The signal-carrying part of each retrieved arm is modeled as a lossy Gaussian channel, \displaystyle\hat{u}_{1} \displaystyle=e^{i\phi_{u}}\Bigl[\sqrt{\eta_{\rm mem}(\tau)}\,\hat{u}_{0}+\sqrt{1-\eta_{\rm mem}(\tau)}\,\hat{v}_{u}\Bigr], \displaystyle\hat{\ell}_{1} \displaystyle=e^{i\phi_{\ell}}\Bigl[\sqrt{\eta_{\rm mem}(\tau)}\,\hat{\ell}_{0}+\sqrt{1-\eta_{\rm mem}(\tau)}\,\hat{v}_{\ell}\Bigr], (4) where \hat{v}_{u} and \hat{v}_{\ell} are vacuum loss modes. The two memories are assumed to have the same Gaussian write-store-read efficiency 17 ; 1 \eta_{\rm mem}(\tau)=\eta_{0}\exp\!\left[-\left(\frac{\tau}{\tau_{\rm mem}}\right)^{2}\right], (5) where \eta_{0} is the zero-storage efficiency and \tau_{\rm mem} is the Gaussian decay constant. Only the phase difference matters for the readout, and, to the first order, we write \phi\equiv\phi_{u}-\phi_{\ell}=\phi_{b}+\phi_{v}+\delta\phi, (6) where \phi_{b} is the bias phase that sets the operating point, \phi_{v} is the velocity-dependent signal phase from Eq. ( 1 ), and \delta\phi is the residual phase fluctuation. The upper memory \mathrm{QM}_{U} is the signal arm and the lower memory \mathrm{QM}_{L} is the reference. Thus \phi_{v} is the differential phase generated by the motion of \mathrm{QM}_{U} relative to the stationary reference memory \mathrm{QM}_{L} . After the second beam splitter the dark output mode is \hat{d}=\frac{\hat{u}_{1}-\hat{\ell}_{1}}{\sqrt{2}}. (7) Linearizing around the dark fringe, \phi_{b}=0 and |\phi_{v}+\delta\phi|\ll 1 , gives \hat{d}\simeq\sqrt{\eta(\tau)}\,\hat{b}_{0}+i\sqrt{\eta(\tau)N}\,(\phi_{v}+\delta\phi)+\hat{d}_{\rm vac}+\hat{d}_{n}, (8) where \eta(\tau)=\eta_{\rm ext}\eta_{\rm mem}(\tau)=\eta_{\rm ext}\eta_{0}\exp\!\left[-\left(\frac{\tau}{\tau_{\rm mem}}\right)^{2}\right] (9) is the total signal transmission to the homodyne

Read the complete text