Controlling Outlier Contamination In Multimessenger Time-domain Searches For Supermasssive Binary Black Holes

TL;DR

This paper introduces a Gibbs sampling method for robust outlier detection in multimessenger time-domain data, improving signal searches for supermassive binary black holes with low computational costs.

Contribution

It proposes a simple, adaptable Gibbs sampling scheme for outlier mitigation in time-series data relevant to supermassive binary black hole searches, applicable to pulsar-timing and photometric datasets.

Findings

Successfully diagnosed outliers in simulated pulsar datasets.

Applied methods to real pulsar J1909-3744 from NANOGrav data.

Explored outlier effects in binary-AGN candidate PG1302-102.

Abstract

Time-domain datasets of many varieties can be prone to statistical outliers that result from instrumental or astrophysical anomalies. These can impair searches for signals within the time series and lead to biased parameter estimation. Versatile outlier mitigation methods tuned toward multimessenger time-domain searches for supermassive binary black holes have yet to be fully explored. In an effort to perform robust outlier isolation with low computational costs, we propose a Gibbs sampling scheme. This provides structural simplicity to outlier modeling and isolation, as it requires minimal modifications to adapt to time-domain modeling scenarios with pulsar-timing array or photometric data. We robustly diagnose outliers present in simulated pulsar-timing datasets, and then further apply our methods to pulsar J$1909$$-$$3744$ from the NANOGrav 9-yr Dataset. We also explore the periodic binary-AGN candidate PG$1302$$-$$102$ using datasets from the Catalina Real-time Transient Survey, All-Sky Automated Survey for Supernovae, and the Lincoln Near-Earth Asteroid Research. We present our findings and outline future work that could improve outlier modeling and isolation for multimessenger time-domain searches.