We describe the multi-band template analysis (MBTA) search for gravitational waves signals from coalescences of compact objects in the LIGO–Virgo data, at the time of the third observing run (2019–2020), both for low-latency detections and for offline analysis. Details are given on the architecture and functioning of the pipeline, including transient noise mitigation strategies, parameter space for the searched signals, detection of candidates and evaluation of a false alarm rate associated to them. The performance of the low-latency search is demonstrated based on the LIGO–Virgo third observing run, during which MBTA has contributed to 42 alerts, submitting candidates with a median latency of 36s. The performance of the offline search is illustrated on a subset of data collected during the second LIGO–Virgo observation run in 2017, and are quantified based on injections of simulated signal events on the same data.

我们描述了在第三次观测运行（2019-2020 年）时，多波段模板分析 (MBTA) 搜索来自 LIGO-Virgo 数据中致密物体聚结的引力波信号，用于低延迟检测和离线分析。详细介绍了管道的架构和功能，包括瞬态噪声缓解策略、搜索信号的参数空间、候选信号的检测以及与其相关的误报率的评估。低延迟搜索的性能基于 LIGO-Virgo 第三次观测运行，在此期间 MBTA 已贡献了 42 个警报，提交的候选者的中位延迟为 36 秒。离线搜索的性能在 2017 年第二次 LIGO-Virgo 观测期间收集的数据子集上进行了说明，