Hintze Centre for Astrophysical Surveys

Abstract:

Modern astronomical surveys deliver immense volumes of transient detections, yet distinguishing real astrophysical signals (for example, explosive events) from bogus imaging artefacts remains a challenge. Convolutional neural networks are effectively used for real versus bogus classification; however, their reliance on opaque latent representations hinders interpretability. Here we show that large language models (LLMs) can approach the performance level of a convolutional neural network on three optical transient survey datasets (Pan-STARRS, MeerLICHT and ATLAS) while simultaneously producing direct, human-readable descriptions for every candidate. Using only 15 examples and concise instructions, Google’s LLM, Gemini, achieves a 93% average accuracy across datasets that span a range of resolution and pixel scales. We also show that a second LLM can assess the coherence of the output of the first model, enabling iterative refinement by identifying problematic cases. This framework allows users to define the desired classification behaviour through natural language and examples, bypassing traditional training pipelines. Furthermore, by generating textual descriptions of observed features, LLMs enable users to query classifications as if navigating an annotated catalogue, rather than deciphering abstract latent spaces. As next-generation telescopes and surveys further increase the amount of data available, LLM-based classification could help bridge the gap between automated detection and transparent, human-level understanding.

Abstract:

Atomic hydrogen constitutes the gas reservoir from which molecular gas and star formation in galaxies emerges. However, the weakness of the line means it has been difficult to directly detect in all but the very local Universe. Here, we present results from the first search using the MeerKAT International Tiered Extragalactic Exploration (MIGHTEE) Survey for high-redshift () H i emission from individual galaxies. By searching for 21-cm emission centred on the position and redshift of optically selected emission-line galaxies we overcome difficulties that hinder untargeted searches. We detect 11 galaxies at , forming the first sample of detections with an interferometer, with the highest redshift detection at . We find they have much larger H i masses than their low-redshift H i-selected counterparts for a given stellar mass. This can be explained by the much larger cosmological volume probed at these high redshifts, and does not require any evolution of the H i mass function. We make the first-ever measurement of the baryonic Tully–Fisher relation (bTFr) with H i at and find consistency with the local bTFr, but with tentative evidence of a flattening in the relation at these redshifts for higher-mass objects. This may signify evolution, in line with predictions from hydrodynamic simulations, or that the molecular gas mass in these high-mass galaxies could be significant. This study paves the way for future studies of H i beyond the local Universe, using both searches targeted at known objects and via pure H i selection.