Multiparameter estimation for the superresolution of two incoherent sources
Antonin Grateau 1, Alexander Boeschoten 1, Tanguy Favin-Lévêque 1, Isael Herrera 1, Nicolas Treps 1 · 2026
Read Multiparameter estimation for the superresolution of two incoherent sources by Antonin Grateau 1, Alexander Boeschoten 1, Tanguy Favin-Lévêque 1, Isael Herrera 1, Nicolas Treps 1 (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-2601-14876.txt.
Introduction
Antonin Grateau, Alexander Boeschoten, Tanguy Favin-Lévêque, Isael Herrera, and Nicolas Treps test a way to see two light sources that sit very close together. Their 2026 arXiv paper is about beating the blur limit of normal optics. The two sources shine on their own, with no shared phase.
The team sorts the light by its spatial modes, a method called SPADE. They use two such sorters, with one shifted on purpose. In the lab they gauge the gap, the center point, and the brightness split at once. The gaps they find sit well below the classic blur limit.
Catalog record
| CREATOR | Antonin Grateau 1, Alexander Boeschoten 1, Tanguy Favin-Lévêque 1, Isael Herrera 1, Nicolas Treps 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-14876
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Multiparameter estimation for the superresolution of two incoherent sources Preprint: APS/123-QED Antonin Grateau 1, Alexander Boeschoten 1, Tanguy Favin-Lévêque 1, Isael Herrera 1, Nicolas Treps 1 Abstract We experimentally demonstrate the simultaneous estimation of the three parameters characterizing a pair of incoherent optical sources in the sub-Rayleigh regime, enabling super-resolved scene characterization. Using spatial-mode demultiplexing (SPADE) with two demultiplexers—one deliberately shifted—we determine separations well below the diffraction limit and achieve sensitive joint estimation of separation, centroid, and relative brightness over a broad range of scene configurations in a single experimental setting. We benchmark our performance using Fisher-information-based Cramér–Rao bounds, and discuss the corresponding quantum limits. We investigate two complementary scenarios: a realistic case with slightly non-identical sources, and an idealized case of indistinguishable sources. I Introduction Imaging is the process of extracting information about a system or object from the light that it emits or scatters. It can therefore be formulated as a parameter estimation problem: one or more parameters – for example, the object’s position, its size, and its brightness – are encoded in the light, the carrier of information, and need to be estimated 16 ; 17 ; 15 ; 5 ; 3 ; 11 ; 10 ; 30 . The detection of the light and estimation strategy of the parameter from the data constitute the measurement and determine the quality of the estimation and its fundamental limitations. Conventionally, imaging employs a pixelized intensity measurement, also known as direct imaging (DI). However, for certain imaging tasks, DI has severe limitations and other measurement strategies should be employed. An important example is the estimation of the separation between two incoherent point sources, a single-parameter estimation problem. DI is limited by diffraction of the light on a finite aperture, and therefore unable to resolve objects when their overlap is considerably larger than the size of the point-spread function 23 ; 1 ; 31 ; 29 . Evaluating this problem from a quantum metrology point of view, it becomes clear that this is not a fundamental limitation. In particular, Tsang et al. 35 showed that whereas in DI the information per photon vanishes at small separations, the quantum Cramér–Rao bound (QCRB), which sets the ultimate precision limit, remains finite and separation independent, even in the limit of zero separation. It was shown that an intensity measurement in the Hermite–Gauss (HG) mode basis, known as spatial mode demultiplexing (SPADE), achieves this bound. This technique has been experimentally implemented, and the superiority of SPADE over DI is verified in several experiments using different techniques 26 ; 22 ; 33 ; 37 ; 34 ; 38 ; 14 ; 28 . However, realistic imaging scenarios are inherently multiparametric. For instance, the image of two point sources is defined by their separation, their centroid position, and their relative intensities. These parameters are simultaneously encoded in the optical field and are not independent 25 ; 24 . Therefore, the ultimate precision limits and the optimal measurement strategies require the framework of multiparameter estimation theory 19 ; 7 ; 4 . From this perspective, multiparameter estimation can be viewed as the step between single-parameter estimation and full image reconstruction, thereby enabling super-resolved imaging of simple scenes 6 ; 12 ; 13 ; 18 ; 27 ; 9 ; 8 . The joint estimation of centroid and distance has been performed with different techniques, all based on SPADE 32 ; 21 . However, these experiments still assume the sources to have equal intensity. Recently, the simultaneous estimation of the intensity imbalance and separation, assuming a known barycenter, was performed 36 , but the full three parameter estimation has only been done in the time domain and sequentially 2 . Indeed, the real challenge of multiparameter estimation lies not only in achieving high precision for a specific scene, but also in designing a measurement strategy that provides robust sensitivity over a wide range of possible scenes, without prior knowledge of the parameters to be estimated. In this letter, we experimentally perform the simultaneous estimation of the separation, centroid, and intensity imbalance of two incoherent point sources over a broad range of parameter values. This is done in a single experimental setting, without adapting the measurement to the unknown scene. Our approach relies on two spatial demultiplexers (Multi-Plane Light Converters, MPLCs), with one deliberately shifted to enhance the information. This scheme is more robust and versatile than implementing the full set of optimal modes 24 ; 2 , which are at the same time scene-dependent and challenging to realize in practice, and thus experimentally relevant. Importantly, our analysis highlights the subtleties that arise when moving from the ideal theoretical case to realistic experimental conditions, where the two sources are slightly different, and paves the way for practical applications. In what follows, we first show how the dual-MPLC configuration enhances parameter sensitivity compared to a single-MPLC setup. We then present results for experimental source configurations, and finally address the more fundamentally challenging scenario of indistinguishable sources, investigated using data obtained from a single light source. II Two mode-basis approach for multiparameter estimation Performing multiparameter estimation is not only about achieving good precision for a specific scene, but also about designing an experimental configuration that provides robust sensitivity over a wide range of possible scenes. In this work, rather than tailoring an optimal mode decomposition for a particular parameter set, we aim at a versatile measurement strategy that remains informative across the full parameter space. To this end, we introduce a dual-basis approach based on two commercial multi-plane light converters (MPLCs), which decompose the incoming field into two shifted Hermite–Gaussian mode bases. It is well known that a single Hermite–Gaussian basis is insufficient to simultaneously estimate the source separation d , centroid c , and brightness imbalance p , due to the intrinsic symmetries of this basis. Previous work has shown that the unavoidable asymmetries of a practical MPLC can be exploited to access limited multiparameter information for equally bright sources 32 . However, this effect is weak and does not enable three-parameter estimation. We therefore use two mode bases: the first MPLC provides a practical Hermite–Gaussian mode decomposition, while the second MPLC is deliberately shifted laterally, resulting in a complementary mode basis. In practice, we use the first four horizontal modes of the first MPLC to extract most of the available information, and we include a single mode from the shifted MPLC to lift remaining degeneracies. This addition significantly improves robustness, reducing estimator bias and enhancing sensitivity across the parameter space. To quantify the benefit of the dual-MPLC configuration, we compare its theoretical sensitivity to that of a single-MPLC setup using experimentally realistic conditions, including imperfect mode decomposition and source non-idealities. We construct estimators directly from experimentally measured calibration curves. By using either one or two sets of calibration curves, we can model distinguishable and indistinguishable sources, respectively. For each configuration, we compute the classical Fisher information matrix and the associated Cramér–Rao bounds (CRBs), assuming shot-noise-limited detection (see Supp. Mat.). These bounds provide a lower limit on the achievable standard deviation for each parameter \alpha\in\{d,c,p\} , \sigma_{\alpha}\geq\sigma_{\alpha,\mathrm{CRB}}=\sqrt{(\mathcal{F}^{-1})_{\alpha\alpha}}. (1) We perform our simulations on an ensemble of scenes spanning the relevant parameter range at fixed total photon number. For each configuration (one