The NANOGrav 15-year data set: high-precision timing of 68 millisecond pulsars

The North American Nanohertz Observatory for Gravitational Waves (NANOGrav) Physics Frontiers Center monitors a growing array of Millisecond pulsars (MSPs) and conducts high-precision timing over decades-long timescales with the aim of directly detecting gravitational waves (GWs) from merging supermassive black hole binary systems. GWs passing through our galaxy induce correlated fluctuations in the observed pulse arrival times from MSPs; monitoring a large (≳50) set of pulsars with sub-microsecond precision is necessary for GW detection. The NANOGrav observing program currently monitors a set of 72 MSPs using the Green Bank Telescope, the Very Large Array, and the CHIME Telescope; additional pulsars are observed with CHIME and FAST, for exploratory purposes. Observational results are organized around a set of data releases; every few years, data from all pulsars are compiled, reduced, and used as the basis for a new set of GW analyses. Described here is the latest NANOGrav “15-year” data set, including observations of 68 MSPs with the GBT, Arecibo, and VLA, ending August 10, 2020; this end date marks NANOGrav’s final observations at Arecibo Observatory, which collapsed in December of 2020. Data were analyzed with a new, entirely version-controlled pipeline to generate narrowband and wideband pulse times of arrival (TOAs; toagen), identify and excise corrupted data, and establish high-precision narrowband and wideband timing models for all pulsars included, employing PINT timing software (timing_analysis). This pipeline, composed of toagen and timing_analysis repositories, is written primarily in python, uses Jupyter notebooks to standardize processing steps, and configuration (.yaml) files to facilitate reproducibility and record decisions made while modeling individual sources; it will be released along with the 15-year data set in 2022. This analysis incorporates VLA data for the first time and represents the largest jump in the number of MSPs in any NANOGrav data set (21).