The recent direct observation of gravitational waves has further emphasized the desire for fast, low-cost, and accurate methods to infer the parameters of gravitational wave sources. Due to expense in waveform generation and data handling, the cost of evaluating the likelihood function limits the computational performance of these calculations. Building on recently developed surrogate models and a novel parameter estimation pipeline, we show how to quickly generate the likelihood function as an analytic, closed-form expression. Using a straightforward variant of a production-scale parameter estimation code, we demonstrate our method using surrogate models of effective-one-body and numerical relativity waveforms. Our study is the first time these models have been used for parameter estimation and one of the first ever parameter estimation calculations with multi-modal numerical relativity waveforms, which include all l <= 4 modes. Our grid-free method enables rapid parameter estimation for any waveform with a suitable reduced-order model. The methods described in this paper may also find use in other data analysis studies, such as vetting coincident events or the computation of the coalescing-compact-binary detection statistic.
Surrogate model for gravitational wave signals from non-spinning, comparable- to large-mass-ratio black hole binaries built on black hole perturbation theory waveforms calibrated to numerical relativity
Architecture Acceleration of Machine Learning Gravitational Waveform Surrogate Models
INFN, Sezione di Firenze(意大利国家核物理研究所佛罗伦萨分部);INFN, Sezione di Pisa(意大利国家核物理研究所比萨分部);INAF, Osservatorio di Astrofisica e Scienza dello Spazio(意大利国家天体物理研究所空间科学与天体物理观测站);INFN, Sezione di Bologna(意大利国家核物理研究所博洛尼亚分部)
Fast prediction and evaluation of gravitational waveforms using surrogate models
[摘要]我们提出了一种解决方案,可以快速准确地预测任何给定物理模型中的引力波形。该方法与实时应用和使用标准方法生成波形可能过于昂贵的更传统的场景都相关。我们的方法是基于三个离线步骤,从而在一个准确的降阶模型,可以用作替代品的真实/基准波形家庭。首先,使用贪婪算法确定一组m个参数值,从该贪婪算法构造减少的基表示。第二,这m个参数导致选择m个时间值,用于使用经验内插器来内插波形时间序列。第三,在这m个时间中的每一个时间对波形的值执行参数维度中的拟合。对于一般参数选择,预测L个波形时间样本的成本是m L + m c_f在线操作的数量级,其中c_f表示拟合函数操作计数,并且通常m << L。我们生成准确的替代模型的有效一体(EOB)波形的非旋转的二元黑洞合并的持续时间长达10^5 M,质量比从1到10,和多个谐波模式。我们发现,这些替代品是三个数量级更快地评估相比,以标准的方式生成EOB波形的成本。其他波形模型的替代模型构建遵循相同的步骤,并且具有相同的低在线缩放成本。对于昂贵的数值模拟的二元黑洞coalescences,我们因此预计大的加速产生新的波形与代理。由于波形生成是参数估计算法和参数空间探索的主要成本之一,因此代理模型提供了一种新的实用方法,可以在不影响精度的情况下显着加速此类研究。
Chase Orbits, not Time: A Scalable Paradigm for Long-Duration Eccentric Gravitational-Wave Surrogates
International Centre for Theoretical Sciences, Tata Institute of Fundamental Research(国际理论科学中心,塔塔基础研究所);University of Massachusetts, Dartmouth(马萨诸塞大学达特茅斯分校);Indian Institute of Technology Madras(印度马德拉斯理工学院);Max Planck Institute for Gravitational Physics (Albert Einstein Institute)(马克斯·普朗克引力物理研究所(阿尔伯特·爱因斯坦研究所))
Discovery of Interpretable Surrogates via Agentic AI: Application to Gravitational Waves
Kavli Institute for Theoretical Physics, University of California Santa Barbara(加州大学圣芭芭拉分校凯文利理论物理研究所);Center for Gravitational Physics, University of Texas at Austin(德克萨斯大学奥斯汀分校重力物理中心);Department of Physics, University of California at Santa Barbara(加州大学圣芭芭拉分校物理系);International Centre for Theoretical Sciences, Tata Institute of Fundamental Research(塔塔基础研究机构国际理论科学研究中心);School of Natural Sciences, Institute for Advanced Study(高级研究院自然科学学院);Chennai Mathematical Institute(钦奈数学研究所);Kavli Institute for Cosmological Physics, The University of Chicago(芝加哥大学凯文利宇宙物理研究所);Department of Particle Physics & Astrophysics, Weizmann Institute of Science(魏茨曼科学研究所粒子物理与天体物理系)
An automated parameter domain decomposition approach for gravitational wave surrogates using hp-greedy refinement
Facultad de Matemática, Astronomía, Física y Computación, Universidad Nacional de Córdoba(科尔多瓦国立大学数学、天文学、物理学与计算机科学学院);CONICET(国家科学技术研究委员会);ISISTAN-CONICET Research Institute, UNICEN University(阿根廷中部国立大学ISISTAN-CONICET研究所)
Deep Residual Error and Bag-of-Tricks Learning for Gravitational Wave Surrogate Modeling
Aristotle University of Thessaloniki(塞萨洛尼基亚里士多德大学);Department of Informatics, Aristotle University of Thessaloniki(塞萨洛尼基亚里士多德大学信息学系);Department of Physics, Aristotle University of Thessaloniki(塞萨洛尼基亚里士多德大学物理系);GSI Helmholtz Center for Heavy Ion Research(亥姆霍兹重离子研究中心(GSI))
Neural surrogates for designing gravitational wave detectors
Max Planck Institute for the Science of Light(马克斯·普朗克光科学研究所);ELLIS Unit Linz, Institute for Machine Learning(ELLIS林茨单元,机器学习研究所);Machine Learning in Science Cluster, Department of Computer Science, Faculty of Science, University of Tuebingen(科学机器学习集群,计算机科学系,科学学院,图宾根大学);LIGO Laboratory, California Institute of Technology(LIGO实验室,加州理工学院);School of Physics and Astronomy, Tel Aviv University(物理与天文学学院,特拉维夫大学)
Gravitational wave surrogate model for spinning, intermediate mass ratio binaries based on perturbation theory and numerical relativity
University of Massachusetts, Dartmouth(达特茅斯马萨诸塞大学);University of Texas at Austin(德克萨斯大学奥斯汀分校);University of Rhode Island(罗德岛大学);University of California Santa Barbara(加州大学圣塔芭芭拉分校);University of Virginia(弗吉尼亚大学);Case Western Reserve University(凯斯西储大学)
Gravitational wave memory of the binary black hole events in GWTC-2
引力波记忆是广义相对论的一个重要预言。现有的GW内存检测工作主要集中在波形分析上。由于GW存储器的准直流特性和弱点,波形分析方法很难检测到GW存储器。在这项工作中,我们实现了一个完全不同的方案来估计GW内存。在该方案中,我们首先基于数值相对论模拟,采用Bondi-Metzner-Sachs方法计算了双星黑洞的GW记忆。然后,我们构造了一个代理模型来关联二元黑洞的参数和GW记忆。之后,我们应用这个代理模型与贝叶斯技术,估计GW的记忆的48个二元黑洞事件记录在GWTC-2。已经估计了与所有48个事件相对应的GW存储器。最有趣的结果是GW 190814。对于Hanford检测器和Livingston检测器,相应的GW存储器分别约为$-1\times10^{-23}$和$1\times10^{-23}$。同时,我们发现3个$σ$ C. L。GW 190814的记忆菌株在Hanford检测器上为阴性,而在Livingston检测器上为阳性。
Eccentric binary black holes: A new framework for numerical relativity waveform surrogates
Max Planck Institute for Gravitational Physics (Albert Einstein Institute)(马克斯·普朗克引力物理研究所(爱因斯坦研究所));Department of Mathematics, Center for Scientific Computing and Data Science Research, University of Massachusetts, Dartmouth(马萨诸塞大学达特茅斯分校数学系、科学计算与数据科学研究中心);Kavli Institute for Theoretical Physics, University of California Santa Barbara(加州大学圣塔芭芭拉分校卡弗里理论物理研究所);Cornell Center for Astrophysics and Planetary Science, Cornell University(康奈尔大学康奈尔天体物理学与行星科学中心);Nicholas and Lee Begovich Center for Gravitational-Wave Physics and Astronomy, California State University Fullerton(加州州立大学富勒顿分校尼古拉斯和莉·贝戈维奇引力波物理与天文学中心);International Centre for Theoretical Sciences, Tata Institute of Fundamental Research(塔塔基础研究院国际理论科学中心);Departament de Física, Universitat de les Illes Balears, IAC3 – IEEC(巴利阿里群岛大学物理系,IAC3–IEEC);Department of Physics and Weinberg Institute for Theoretical Physics, University of Texas at Austin(德克萨斯大学奥斯汀分校物理系与温伯格理论物理研究所);CENTRA, Departamento de Física, Instituto Superior Técnico, Universidade de Lisboa(里斯本大学高等理工学院物理系CENTRA中心);Theoretical Astrophysics 350-17, California Institute of Technology(加州理工学院理论天体物理学部);Department of Physics, Vivekananda Satavarshiki Mahavidyalaya (affiliated to Vidyasagar University)(维韦卡南达萨瓦沙基学院物理系(隶属于维迪亚萨加尔大学))
NRSur7dq4v2: A multi-domain precessing surrogate model with improved accuracy
University of Massachusetts, Dartmouth(麻省大学达特茅斯分校);Cornell University(康奈尔大学);The University of Mississippi(密西西比大学);California Institute of Technology(加州理工学院)
AI and extreme scale computing to learn and infer the physics of higher order gravitational wave modes of quasi-circular, spinning, non-precessing binary black hole mergers
Argonne National Laboratory(阿贡国家实验室);University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);University of Chicago(芝加哥大学);Tata Institute of Fundamental Research(塔塔基础研究所)
Hyperparameter optimization of hp-greedy reduced basis for gravitational wave surrogates
CONICET(阿根廷国家科学与技术研究理事会);Facultad de Matemática, Astronomía, Física y Computación, Universidad Nacional de Córdoba(科尔多瓦国立大学数学、天文、物理与计算机学院);ISISTAN-CONICET Research Institute(ISISTAN-阿根廷国家科学与技术研究理事会研究所);UNICEN University(中阿根廷国立大学);Facultad de Ciencias Exactas y Tecnología, Universidad Nacional de Tucumán(图库曼国立大学精确科学与技术学院)
Merger remnant and eccentricity dynamics surrogates for eccentric nonspinning black hole binaries
University of Massachusetts, Dartmouth(麻省大学达特茅斯分校);Max Planck Institute for Gravitational Physics(Albert Einstein Institute)(马克斯·普朗克引力物理研究所(爱因斯坦研究所));Cornell University(康奈尔大学);University of California Santa Barbara(加州大学圣塔芭芭拉分校);California State University Fullerton(加州州立大学富勒顿分校);Northwestern University(西北大学)