arXiv Science⌕ Search
Home/Archive/arXiv

arXiv · 2610.33942

O Mirror, Where Art Thou?

Sai Krishanth P. M.·Sanchit Sabhlok·Jaren N. Ashcraft·Kyle Van Gorkom·Ewan S. Douglas
View original source ↗Read source document ↗Export metadata ↓

Abstract

Wavefront sensing reveals that a telescope is misaligned, but not which mirror moved or by how much; inferring the mirror positions themselves turns alignment into a direct correction. We describe a simulation framework for inferring telescope mirror misalignment from multi-field Point Spread Function (PSF) measurements projected onto Circular-Bessel and Zernike basis vectors. Two case studies are presented: a Hubble Space Telescope model with five secondary-mirror degrees of freedom and a three-mirror anastigmat with ten primary and secondary mirror degrees of freedom. In both cases, field-dependent multivariate Gaussian likelihoods are explored with nested sampling and refined with differential evolution. Recovered states are validated with residual PSF images and optical path difference (OPD) measurements. In the Hubble benchmark, differential evolution recovers injected misalignments to less than 0.01 microns and reduces on-axis OPD error from 2.3 waves to 3 x 10E-8 waves. In the TMA study, the posterior shows broad marginals and correlated parameter pairs that suggest partial degeneracies at higher dimension; differential evolution recovers the truth alignment state less precisely than in the Hubble benchmark (less than 17 microns) and reduces on-axis OPD error from 8.24 waves to 2 x 10E-3 waves. This generalizable algorithm enables direct sensing of the state of an optical system and automated alignment.

Explore related subjects

astro-ph.IM

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Sai Krishanth P. M., Sanchit Sabhlok, Jaren N. Ashcraft, Kyle Van Gorkom, Ewan S. Douglas. 2026-10-24. O Mirror, Where Art Thou?. https://arxiv.org/abs/2610.33942

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

KEEP EXPLORING

Related papers

arXiv2026

spherical: A Comprehensive Database and Automated Pipeline for VLT/SPHERE High-Contrast Imaging

I present spherical (https://github.com/m-samland/spherical), a software package and database for the ESO VLT/SPHERE high-contrast imager. SPHERE has produced the world's largest archive of direct imaging observations of exoplanets, substellar companions, and circumstellar disks, but the archive can only be searched by header keywords, and raw data and calibrations must be retrieved by hand. spherical combines a curated, searchable database of the complete SPHERE observation history with automated reduction pipelines for the Integral Field Spectrograph (IFS) and for IRDIS dual-band imaging (DBI). Observations are cross-matched against SIMBAD, Gaia DR3, and the MOCA database, so samples can be selected by stellar properties, age, and association membership as well as by observing mode and conditions. The August 2026 release, archived on Zenodo (https://doi.org/10.5281/zenodo.15147730), covers 2014 May to 2026 August and includes 6101 IRDIS DBI, 4733 IFS, 1205 IRDIS polarimetric, and 355 sparse aperture masking sequences. Each release comes with metadata, such that an analysis can be traced to the exact database state it used. The pipelines download raw data and calibrations from the ESO archive, extract IFS spectral cubes with the adapted open-source CHARIS pipeline, calibrate IRDIS data in a parallel workflow, and post-process both with the TRAP algorithm for companion detection, astrometry, and contrast spectra. The code is written in Python and installs with pip, without ESO pipeline software, and the IRDIS reduction reproduces the published photometry and astrometry of 51 Eridani b. spherical makes it practical to assemble homogeneous samples for population studies and to reprocess the full archive uniformly, and it can also serve as the retrieval and pre-processing stage for community tools such as VIP, pyKLIP, and IRDAP.

astro-ph.IM↗
arXiv2026

Emulation of non-linear 1D spectral models: relativistic X-ray reflection

The use of machine learning techniques to approximate computationally expensive models has become increasingly prevalent in a wide variety of fields within astronomy. We discuss the implementation of emulators for 1-dimensional models in the context of the astrophysical numerical model reltrans, a black hole X-ray spectral model that models the effects of relativistically smeared emission from an accretion disk. We argue that the decision of whether and how to emulate should follow from a systematic characterisation of the target model, and we demonstrate a diagnostic workflow: examining how the spectrum varies with individual parameters. We adopt a modular strategy, emulating only the relativistically convolved reflection spectrum (1-10% of the total flux) rather than the full model. Using an operator-learning architecture with Fourier feature embeddings and FiLM conditioning, we reproduce the reflection spectrum to O(0.1)% precision across 0.1-100 keV with a 4-10x speed-up that scales considerably better under vectorised evaluation. This emulator, RTFAST2, recovers the true parameters of simulated observations without the systematic posterior biases of our previous work. We conclude that no architecture is universally transferable and bespoke emulators motivated by a model's specific structure are required. The modular approach taken in this work presents a promising strategy for future emulators of numerical models.

astro-ph.IM↗
arXiv2026

Data-driven acquisition offset correction for telescope networks using machine learning and sensitivity analysis

Reliable target acquisition is essential for maximizing observing efficiency, particularly for robotic telescope networks that operate across many sites, instruments, and environmental conditions. Although these facilities generate large volumes of operational and observational data, acquisition procedures often continue to rely on static pointing models and manual adjustments by observatory staff. These approaches can be effective, but they may not fully capture changing behavior associated with telescope configuration, observing conditions, instrument state, or other time-dependent effects. We present a data-driven machine-learning pipeline for automatically predicting and correcting telescope acquisition offsets, demonstrated using archival operational logs from the Las Cumbres Observatory network. Our approach applies both traditional machine-learning methods and deep-learning regression models to learn relationships between measured acquisition offsets and a range of observing, telescope, and instrumentation conditions. By using historical performance data, the pipeline is designed to complement existing pointing models and reduce the need for repeated manual correction. We also present an open-source, Dockerized software service that makes the trained models accessible through a browser-based graphical interface. The service is intended to support straightforward, real-time use by observatory staff while remaining portable across computing environments. Together, the prediction pipeline and deployment framework provide a practical path toward more adaptive, automated, and data-informed telescope acquisition.

astro-ph.IM↗