arXiv:2610.16871
Direct polarisation-state reconstruction with a temporal pointer: accessible states, resolution and optimal coupling
Shane P. Gervais, Lambert Giner
Abstract
Weak measurements are used to amplify small signals, to address foundational questions, and to characterise quantum states without tomography. In direct state reconstruction each amplitude of the state vector is proportional to a single complex weak value, which for a temporal pointer is read from the mean arrival time and the mean frequency of the postselected pulse. Such an experiment requires three quantities to be fixed, the coupling, the angular resolution demanded of the reconstruction, and the set of states over which the method remains usable, and these are commonly set by the available apparatus rather than derived. We derive all three for a Gaussian temporal pointer and an arbitrary postselection state. The states that cannot be reconstructed form a spherical cap centred on the state orthogonal to the postselection, whose geodesic half-angle follows from the largest of three independent thresholds and involves no adjustable parameter. Two states are distinguishable only if they differ by more than a minimal step fixed by the coupling, by the number of incident photons and by the sensitivity of the weak value along the path followed. For an inversion at first order in the coupling the systematic and statistical errors are opposed and fix an optimal coupling that varies as the inverse sixth root of the number of incident photons, whereas an exact inversion at finite coupling removes the systematic error, the remaining bound being one on the two-centroid readout rather than on the inversion, since below an amplitude equal to a quarter of the coupling two states give the same pair of centroids while their profiles differ. These results allow a direct reconstruction experiment to be dimensioned before the apparatus is assembled.
AI Summary
The paper asks how coupling strength, accessible input states and angular resolution constrain direct reconstruction of a pure polarisation qubit using a Gaussian temporal pointer. A polarisation-dependent delay $\tau$ couples the qubit to a pulse of width $\sigma$; temporal and spectral centroids, measured in separate configurations, encode a complex weak value. Exact centroid expressions are derived at arbitrary coupling $x=\tau/\sigma$, with weak-limit reconstruction formulae for arbitrary nonzero postselection components. Most precision and accessibility calculations specialise to diagonal postselection. For centroid-based reconstruction, the proposed minimum postselection amplitude is $\varepsilon_{\min}=\max[\sqrt{N_{\min}/N_{\rm total}},\sqrt{\Gamma t_{\rm acq}/N_{\rm total}},x/4]$, defining an exclusion cap of half-angle $2\arcsin\varepsilon_{\min}$ around the orthogonal state. The terms represent minimum detected counts, dark counts and finite-coupling effects. Along a known state-space path, centroid-noise propagation gives $\Delta\varphi_{\min}=1/[x\sqrt N\,|dW/d\varphi|]$. Balancing squared finite-coupling bias against statistical variance yields $x_{\rm opt}\propto N_{\rm total}^{-1/6}$ and reconstruction error proportional to $N_{\rm total}^{-1/3}$. An algebraic finite-coupling inversion removes this truncation bias and recovers $N_{\rm total}^{-1/2}$ statistical scaling, but has two branches: near orthogonal postselection, distinct states share both centroids, with a turning point at $\varepsilon\simeq x/4$ for linear polarisations. Numerical profile Fisher information and Monte Carlo tests supplement the analytic results. Full pulse profiles can distinguish centroid-degenerate states but require profile-based estimation. The analysis assumes pure inputs, a Gaussian pointer and negligible instrumental broadening; postselection offers no total-resource metrological advantage.
Central claim may not hold: The abstract overstates arbitrary-postselection generality: near orthogonality the weak-regime crossover is approximately ε=|μν|x/2, whereas the reported ε=x/4 assumes balanced postselection. The claimed smallest accessible angular step at the coupling-limited exclusion boundary uses a weak-limit sensitivity that fails there, while the exact temporal-centroid sensitivity vanishes at the nearby fold and the spectral centroid supplies no linear-family sensitivity. The finite-coupling Fisher-information comparison incorrectly retains the original Gaussian variance σ² for the sample centroid even when the postselected temporal variance differs substantially. The exclusion threshold is not parameter-free or an accuracy guarantee, because it includes a chosen minimum event count and permits dark counts as large as the signal despite assuming a small dark-to-signal ratio. The fold discussion assigns x_opt=0.348 to the θ=140° state, although its optimum is approximately 0.043 and 0.348 is the optimum for θ=30°.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 19 pages, 6 figures, 4 tables