Predicting Follow-Up Observations of Galaxy Clusters Using Machine Learning

The eROSITA X-ray telescope, launched in 2019, is predicted to observe roughly 100,000 galaxy clusters. More detailed observations of these clusters using other X-ray telescopes, like the Chandra Observatory, are needed to resolve outstanding questions about galaxy cluster physics. Our ability to follow up eROSITA observations is expensive and limited, therefore objects chosen for follow-up must be chosen with care. To aid in the evaluation of potential follow-ups, we have developed a proof-of-concept algorithm for predicting longer, higher quality, observations based on mock eROSITA observations. We do so making use of the hydrodynamical cosmological simulation Magneticum, simulated eROSITA instrument conditions using SIXTE, and a novel convolutional neural network. We are able to closely predict cluster morphology, remove noise, and capture substructure. We emphasize that predictions are no substitute for real observations and that our predictions are intended to aid to follow-up observations, not to replace them.