Detection of Strongly Lensed Arcs in Galaxy Clusters with Transformers

Strong lensing in galaxy clusters probes properties of dense cores dark matter halos mass, studies the distant universe at flux levels and spatial resolutions otherwise unavailable, constrains cosmological models independently. The next-generation large scale sky imaging surveys are expected to discover thousands cluster-scale strong lenses, which would lead unprecedented opportunities for applying lenses solve astrophysical problems. However, dataset challenges astronomers identify extract signals, particularly strongly lensed arcs, because their complexity variety. Hence, we propose a framework detect contains transformer-based detection algorithm an image simulation algorithm. We embed prior information arcs into training data through then train with simulated images. use trained transformer from real data. Results show that our approach could achieve 99.63 % accuracy rate, 90.32 recall 85.37 precision rate 0.23 false positive images almost all observation Besides, interpretation method, have shown method important embedded Next step, test reliability usability approach, will apply it available observations (e.g., DESI Legacy Imaging Surveys) upcoming large-scale surveys, such as Euclid CSST.