SPLIT: Loosely-coherent inference of long-inspiral gravitational-wave sources for robustness against modeling and data outliers
The SPLIT framework addresses systematic biases in long-duration gravitational-wave source analysis by segmenting data into blocks. It analyzes each block coherently and combines results incoherently to prevent phase errors from accumulating over months or years. This semi-coherent approach significantly reduces systematic biases compared to fully coherent methods, offering better parameter accuracy for LISA sources.