<inline-formula><mml:math><mml:msup><mml:mi>χ</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:math></inline-formula> statistic for the identification of strongly lensed gravitational waves from compact binary coalescences
用于识别致密双星并合强透镜引力波的 χ² 统计量
Gravitational waves (GWs) emanated by stellar-mass compact binary coalescences (CBCs), and lensed by galaxy- or cluster-scale lenses, will produce two or more copies of the GW signal. In the geometric-optics limit, these signals will have similar phase evolution but differing amplitudes. Such lensing signatures are expected to be detected by the end of the LIGO-Virgo-KAGRA Collaboration's fifth observing run. In this work, we propose a novel <inline-formula><mml:math><mml:msubsup><mml:mi>χ</mml:mi><mml:mrow><mml:mi>lens</mml:mi></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:math></inline-formula> statistic to segregate pairs of detected GW events as lensed or unlensed, using templates typically used in GW searches. The statistic is an application of the generalized <inline-formula><mml:math><mml:msup><mml:mi>χ</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:math></inline-formula> discriminator, tailored to probe the similarity (or lack thereof) between the phase evolutions of two CBC signals. We assess the performance of <inline-formula><mml:math><mml:msubsup><mml:mi>χ</mml:mi><mml:mrow><mml:mi>lens</mml:mi></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:math></inline-formula> on an in-part astrophysically motivated dataset of lensed and unlensed CBCs detectable in the fourth observing run, assuming a single LIGO-like detector at design sensitivity. Our results show that <inline-formula><mml:math><mml:msubsup><mml:mi>χ</mml:mi><mml:mrow><mml:mi>lens</mml:mi></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:math></inline-formula> correctly identifies lensed events with efficiencies comparable to existing Bayesian and machine learning methods, while being orders of magnitude faster to evaluate than Bayesian methods. Moreover, the statistical properties of <inline-formula><mml:math><mml:msubsup><mml:mi>χ</mml:mi><mml:mrow><mml:mi>lens</mml:mi></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:math></inline-formula> in stationary Gaussian noise are fully understood, in contrast to machine learning methods. <inline-formula><mml:math><mml:msubsup><mml:mi>χ</mml:mi><mml:mrow><mml:mi>lens</mml:mi></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:math></inline-formula> can, therefore, be used to rapidly and accurately weed out the vast majority of unlensed candidate pairs and identify lensed pairs.
展开 ▾巧妙利用相邻模板空间的正交投影构造 χ²_lens,在毫秒量级内完成检验,且在高斯噪声下拥有解析的统计分布,识别效率媲美后验重叠统计量,但无需昂贵的参数估计。
强透镜引力波识别已发展出多种方法,如基于源参数后验重叠的贝叶斯统计量 Haris+ 2018 和机器学习分类器 Goyal+ 2021,但分别面临计算耗时与可解释性不足的问题。本文提出的 χ²_lens 源自广义统一 χ² 框架 Dhurandhar+ 2017,通过构造与较强信号邻域正交的空间并投影较弱信号,高效检验相位相像性,不仅计算极快,且统计性质清晰。未来可结合天空定位重叠度 Haris+ 2018 以及预期时延与放大率先验 More+ 2022 构建联合判别器,并有望集成至亚阈值搜索管道 Li+ 2023,为快速筛选强透镜候选事件和挖掘弱信号图像提供实用工具。
预印本 2025-02-02 · 刊出 2025-12-24 · 收录 2026-10-12