Abstract:The optimal statistic (OS) is a frequentist estimator for the amplitude and significance of a spatially-correlated signal in pulsar timing array (PTA) data, and it is widely used to search for the gravitational wave background (GWB). However, the OS cannot perfectly distinguish between different spatial correlations. In this paper, we introduce the multiple component optimal statistic (MCOS): a generalization of the OS that allows for multiple correlations to be simultaneously fit to the data. We use simulated data to show that this method more accurately recovers injected spatially correlated signals, and in particular reduces the risk of a false detection of a signal with the wrong spatial correlation. We also demonstrate that this method can be used to recover multiple correlated signals.

Generalized optimal statistic for characterizing multiple correlated signals in pulsar timing arrays