Abstract
Standard methodologies for the extraction of the stochastic gravitational wave background (SGWB) from auto- or cross-correlation of interferometric signals often involve the use of a filter function. The standard optimal filter maximizes the signal-to-noise ratio (SNR) between the total SGWB and the noise. We derive expressions for the optimal filter and SNR in the presence of a target SGWB plus other unwanted components. We also generalize the methodology to the case of template-free reconstruction. The formalism allows to easily perform analyses and forecasts that marginalize over foreground signals, such as the typical ΩGW∝ f2/3 background arising from binary coalescence. We demonstrate the methodology with the LISA mission and discuss possible extensions and domains of application.