arXiv:2609.31049
Computational aberration-retrieval with entangled photons
Baptiste Courme, YoonSeok Baek, Hugo Defienne
Abstract
Adaptive optics (AO) is central to high-resolution microscopy, yet conventional implementations often require repeated correction-and-measurement cycles whose number increases with aberration complexity. This measurement burden can become prohibitive when the available temporal window for correction is short or when photon budgets are constrained. Here, we demonstrate an AO strategy that exploits the spatial correlations of entangled photon pairs to retrieve aberrations directly. By using the second-order correlation function as an aberration-sensitive metric, the required correction can be determined with only a few measurements, without relying on image-based feedback or iterative optimization. This correlation-based approach establishes an efficient framework for AO leveraging quantum light, offering a new route toward quantum-enhanced microscopy in regimes where conventional correction strategies become challenging.
AI Summary
The paper addresses whether spatial correlations of entangled photon pairs can support adaptive-optics aberration retrieval with fewer hardware measurements than iterative sensorless correction. For strongly position-anticorrelated pairs, assuming shift-invariant imaging and sufficiently limited aberration complexity, the sum-coordinate projection of the coincidence function satisfies $C^+\propto|h*h|^2$, independently of the object’s spatial structure. Known pupil-plane phase perturbations provide magnitude constraints for computationally retrieving this effective response and extracting a corrective phase. The method uses 3–5 perturbations in simulations and experiments. Acquisition avoids iterative hardware optimization, but reconstruction uses an iterative phase-retrieval algorithm, supplemented by multiple response estimates and singular-value decomposition to handle noise. Experiments generate approximately 810-nm photon pairs using a 405-nm-pumped, 0.5-mm BBO crystal and measure correlations with a $32\times64$-pixel SPAD array. Five random SLM masks enable correction of imposed defocus/astigmatism and speckle-like aberrations, restoring a concentrated correlation peak. Imaging an insect specimen demonstrates recovery of major spatial features after correction; fine-detail recovery remains incomplete because weaker illumination at the pupil periphery limits phase estimation there. The stated applicability constraints are the need to preserve strong correlations at the object and long acquisition times. Aberrations in the illumination path can weaken those correlations, complicating use in embedded specimens. Each correlation image currently requires over $10^7$ frames and several hours, precluding real-time correction. Faster detectors and brighter sources are proposed routes to shorter acquisitions, rather than demonstrated capabilities of this implementation.
Error found: The claimed absence of iterative optimization needs qualification because reconstruction uses iterative magnitude projections, although repeated hardware optimization is avoided. The five reported measurements are correlation acquisitions taking several hours each, so the experiment establishes fewer hardware settings but not a practical acquisition-time or photon-budget advantage. The coherent-PSF recovery formula incorrectly uses |F(H)| instead of sqrt(|F(H)|) for its Fourier-domain amplitude, while its phase-halving operation remains consistent with H = h*h.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.