Adaptive kernel density estimation proposal in gravitational wave data analysis

The Markov Chain Monte Carlo approach is frequently used within a Bayesian framework to sample the target posterior distribution. Its efficiency strongly depends on proposal distribution build chain. best jump one that closely resembles unknown distribution; therefore, we suggest an adaptive based kernel density estimation (KDE). We group model's parameters according their correlation and KDE already accepted points for each group. update KDE-based until it stabilizes. argue such could be efficient in applications where data volume increasing. tested several astrophysical datasets (IPTA LISA) have shown that, some cases, also helps reduce chains' autocorrelation length. of this reduced case strong correlations between large parameters.