Telescopes don’t make catalogues!

Abstract

Telescopes don’t make catalogues, they make intensity measurements; any preciseexperiment performed with a telescope ought to involve modelling those measurements.People make catalogues, but because a catalogue requires hard decisionsabout calibration and detection, no catalogue can contain all of the information in theraw pixels relevant to most scientific investigations. Here we advocate makingcatalogue-like data outputs that permit investigators to test hypotheses with almost thepower of the original image pixels. The key is to provide users approximations tolikelihood tests against the raw image pixels. We advocate three options, in order ofincreasing difficulty: The first is to define catalogue entries andassociated uncertainties such that the catalogue contains the parameters of an approximatedescription of the image-level likelihood function. The second is to produce aK-catalogue sampling in “catalogue space” that samplesa posterior probability distribution of catalogues given the data. The third is to exposea web service or equivalent that can compute the full image-level likelihood for anyuser-supplied catalogue.