Title:Premerger detection of massive black hole binaries using deep learning

Abstract:Coalescing massive black hole binaries (MBHBs) are one of primary sources for space-based gravitational wave (GW) observations. The mergers of these binaries are expected to give rise to detectable electromagnetic (EM) emissions with a narrow time window. The premerger detection of GW signals is vital for follow-up EM observations. The conventional approach for searching GW signals involves high computational costs. In this study, we present a deep learning model to search for GW signals from MBHBs. Our model is able to process 4.7 days of simulated data within 0.01 seconds and detect GW signals several hours to days before the final merger. The model provides the possibility of the coincident GW and EM detection of MBHBs.