arXiv:2611.00521
Optimal Estimation in Weak Value Amplification with a Position Pointer
Jing-Hui Huang, Jing Zhang, Jian Ge, Lichao Liu, Wang Luo, Xiang-Yun Hu
Abstract
We experimentally evaluate five estimators for weak-value amplification (WVA) using a position pointer to measure birefringence time delays. Two general approaches are considered: measuring the position shift $Δx$ and measuring the probability change $ΔP_w$ of successful postselection. Specifically, five estimators are investigated: CCD-detected fringe shift $Δx$, quadrant photodiode (PDQ) transversal voltage difference (XDIFF), and lock-in amplifier output phase $η$ (for $Δx$ measurement), as well as PDQ total output voltage (SUM) and CCD-detected average photon number $N_r$ (for $ΔP_w$ measurement). The measurement precision is characterized by the Allan deviation over a range of input laser powers $P_0$. Generally, measuring the position shift $Δx$ yields higher precision than measuring the probability change $ΔP_w$. At $P_0=0.5$~mW, corresponding to a laser-power-limited regime, direct measurement of the fringe displacement with the CCD achieves a minimum Allan deviation of $(5.30\pm0.10)\times10^{-3}$~as, approaching the corresponding photon-statistical Cramér--Rao bound (CRB). At $P_0=50$~mW, corresponding to a detector-limited regime, PDQ detection combined with lock-in demodulation provides the best precision, reaching $(1.80\pm0.06)\times10^{-3}$~as and remaining close to the corresponding CRB. These results demonstrate that the optimal readout depends on the available optical power, detector dynamic range, and technical noise, providing practical guidance for selecting readout strategies in high-precision WVA-based metrology.
AI Summary
Which readout gives the best practical precision in weak-value-amplified birefringence-delay measurements? This paper experimentally compares five observables in a position-pointer WVA setup: CCD fringe displacement and integrated photon number, quadrant-photodiode differential and total signals, and lock-in output phase. Opposite postselections in a double-slit interferometer convert amplified longitudinal phase shifts into transverse fringe displacements. Each observable is calibrated against delays introduced by tilting a half-wave plate, using ten independent linear fits; noise recorded at zero applied delay is converted into delay Allan deviations. At input power $P_0=0.5\,\mathrm{mW}$, CCD fringe displacement gives the lowest minimum Allan deviation, $(5.30\pm0.10)\times10^{-3}\,\mathrm{as}$, compared with a calculated photon-statistical Cramér–Rao bound of $3.48\times10^{-3}\,\mathrm{as}$. At $50\,\mathrm{mW}$, photodiode detection with lock-in demodulation performs best, reaching $(1.80\pm0.06)\times10^{-3}\,\mathrm{as}$ against a quoted bound of $1.38\times10^{-3}\,\mathrm{as}$. The reported error bars quantify calibration-coefficient reproducibility. Position-based readouts generally outperform total-power readouts in these measurements. The comparison concerns this apparatus and its power, noise, and acquisition constraints. CCD operation is limited by saturation; total-power signals are sensitive to laser-power fluctuations; and low-power photodiode signals suffer from detector and digitization resolution. Lock-in detection uses a $1\,\mathrm{s}$ time constant, reducing sensitivity to faster variations. The proposed combination of a high-well-depth, high-speed camera with lock-in processing is not demonstrated.
Central claim may not hold: The claimed near-CRB lock-in precision compares a minimum Allan deviation obtained at averaging times above one second with a photon-statistical bound calculated for only 3.33 microseconds, whereas the paper's bound at one second is about 714 times smaller than the reported minimum. The SUM and CCD photon-number delay resolutions at zero delay use finite-range calibration slopes even though the stated postselection-probability model has zero local sensitivity there, invalidating their interpretation as local probability-based delay resolutions.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 5 pages, 3 figures