Some detection tests for low complexity data models and unknown background distribution

TL;DR

This paper develops detection strategies for low complexity signals in scenarios with unknown background noise, applying them to astrophysical problems like exoplanet detection and galaxy identification.

Contribution

It introduces new detection methods that do not require known background distributions and are applicable to astrophysical data analysis.

Findings

Effective detection of exoplanets from radial velocity data

Successful identification of distant galaxies in hyperspectral datacubes

Detection strategies adaptable to unknown background noise

Abstract

We consider several detection situations where, under the alternative hypothesis, the signal admits a low complexity model and, under both the null and the alternative hypotheses, the distribution of the background noise is {unknown}. We present several detection strategies for such cases, whose design relies on exogenous or on endogenous data. These testing procedures have been inspired by and are applied to two specific problems in Astrophysics, namely the detection of exoplanets from radial velocity curves and of distant galaxies in hyperspectral datacubes.