• Bayesian inference
• gravitational waves
• LISA
• supermassive black holes

Galactic models suggest that galaxy mergers lead to the formation of Super Massive Black Hole Binaries (SMBHBs), which evolve through environmental interactions and eventually emit gravitational waves (GWs), merging in the mHz frequency range. While current GW detectors lack sensitivity in this band, the space-based observatory LISA, scheduled for launch in the mid-2030s, will enable direct detection.
This study evaluates the feasibility of analyzing multiple SMBHBs in LISA data using a frequency-domain Bayesian approach. We simulate LISA-like SMBHB signals, incorporating downsampling techniques to optimize computational efficiency, and perform Bayesian inference. Our results confirm accurate single-source reconstruction but reveal challenges in multi-source analysis, including high computational costs and biases in individual signal recovery due to residual correlations. We demonstrate that downsampling introduces correlation artifacts; by abandoning this technique, we mitigate some issues for non-overlapping events but find that correlated sources remain indistinguishable as separate signals.
Furthermore, we investigate LISA’s capability to detect hierarchical triple systems via Kozai-Lidov oscillations, demonstrating its potential to probe SMBHB formation channels.
These findings highlight both the opportunities and challenges of SMBHB reconstruction in LISA data analysis, emphasizing the need for improved techniques to enhance source separation and computational efficiency.