arXiv Science⌕ Search
Home/Archive/arXiv

arXiv · 2610.12051

On-chip filled array hyperspectral kinetic inductance detectors (FHYKID) for line intensity mapping

Emile Prêle·Martino Calvo·Usasi Chowdhury·Gregory Garde·Florence Levy-Bertrand·Alessandro Monfardini
View original source ↗Read source document ↗Export metadata ↓

Abstract

The next generation Line Intensity Mapping (LIM) experiments will require very high throughput instrumentation, and as a consequence high pixel counts focal planes. For the study of the [CII] line in the redshift range z=7-8, we develop a hyperspectral filled focal plane (FHYKID) operating in the range 210-235 GHz and with a spectral resolution reaching 120. The spectral filtering is achieved thanks to planar Slot Split Ring Resonators (SSRR) coupled to Lumped Element Kinetic Inductance Detectors (LEKID). This robust concept achieves, thanks to its simplicity, large quantum efficiency and is scalable, even with current readout electronics systems, to several tens of thousands beams on Sky. Our FHYKID will be one of the four channels of the NIKONA instrument, currently seeking funding and to be installed at the IRAM 30-meter telescope at Pico Veleta. Among the science goals, simultaneously mapping the local (galactic and nearby-galaxies) CO line and the cosmological [CII].

Explore related subjects

astro-ph.IM

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Emile Prêle, Martino Calvo, Usasi Chowdhury, Gregory Garde, Florence Levy-Bertrand, Alessandro Monfardini. 2026-10-08. On-chip filled array hyperspectral kinetic inductance detectors (FHYKID) for line intensity mapping. https://doi.org/10.1117/12.3103640

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

radio-astro-tools: linking radio astronomical data to the astronomical Python ecosystem

We present the radio-astro-tools code suite, which consists of several Python packages that enable analysis of radio data, especially interferometric spectral cubes, in the context of the Astropy software ecosystem. While these tools were designed with radio data in mind, they are built to be general and have applications on data sets at other wavelengths. The core package, spectral-cube, handles reading, writing and analysis of cube data, and it enables straightforward parallelization via dask and joblib backends. Support packages include casa-formats-io and radio-beam, which handle reading of CASA tables & images and reading and manipulation of point spread functions, respectively. The pvextractor package facilitates creation of position-velocity diagrams. The uvcombine package implements the "feather" algorithm for combining single-dish and interferometric data. The development of radio-astro-tools included several contributions to other repositories, including matplotlib, astropy, and regions to support interaction between CASA and other parts of the astronomy software ecosystem. We also present a detailed set of tutorials written in Jupyter notebooks that can be run interactively in a browser, providing more accessible access to data analysis for radio spectral-line data cubes.

astro-ph.IM↗
arXiv2026

Pseudo Closed-Loop Bootstrapping for Reinforcement Learning Based AO Control

High contrast imaging with ground-based telescopes requires extremely precise wavefront control. The performance of the controller ultimately depends on the quality and efficiency of the wavefront sensing technique, which has led to the development of ever more sensitive wavefront sensors (WFS). However, the increase in sensitivity comes at the cost of greater nonlinearity, which poses challenges for conventional linear reconstructors and controllers. Reinforcement Learning (RL) is a branch of machine learning in which control policies are learned by interacting with the environment. RL has recently attracted interest in the field of XAO, and previous studies have demonstrated that a model-based RL variant, the Policy Optimization for Adaptive Optics (PO4AO), can effectively compensate for temporal delays, misregistration errors, and moderate WFS nonlinearities. PO4AO learns an initial control strategy by collecting closed-loop data from the integrator controller with a linear reconstructor. However, in cases of highly non-linear WFS and challenging conditions, closing the loop with the integrator is difficult, and high-fidelity data cannot be collected. Consequently, learning robust control with PO4AO is challenging. We propose a robust learning strategy in which PO4AO is pre-calibrated using an internal light source and a DM, enabling direct on-sky closed-loop operation without a stable integrator controller for bootstrapping the initial models.

astro-ph.IM↗
arXiv2026

A Coherent Harmonic Summing Pulsar Search Code

Pulsars have narrow pulses, so their signals spread over many harmonics in a Fourier transform. Standard pulsar searches add the powers of those harmonics and discard their phases, which throws away the pulse shape and some of the signal. Folding the time series at each trial period keeps both, but is expensive even with the Fast Folding Algorithm (FFA). We describe CoherentSearch.jl, an open-source Julia code that reconstructs the pulse profile at every trial spin frequency directly from the Fourier transform. It interpolates the complex amplitudes of typically 60 harmonics, inverse transforms them into a profile, and tests that profile against boxcar templates of many widths. Because the input spectrum is normalized, the noise in every profile is known in closed form, so one threshold gives one false-alarm rate across the whole search. In an injection study of over 860,000 simulated pulsars at a matched false-alarm rate, the new search recovers 76% of the pulsars in white noise, against 64% for riptide's FFA in its recommended configuration and 42% for PRESTO's accelsearch, and its threshold does not move under red noise. On one CPU core it is 1.4-2.2 times faster than riptide at matched frequency coverage, it scales to 27 times faster on 48 cores, and modern GPUs run it up to nine times faster than a 20-core workstation.

astro-ph.IM↗