Development of Efficient Black Hole Spectroscopy and a Desktop Cluster for Detecting Compact Binary Mergers

该奖项支持两个项目。第一个任务是开发在宇宙中最极端的环境之一——黑洞视界附近——测试爱因斯坦相对论的方法。爱因斯坦的理论预测，黑洞发出的引力波应该由特定的频率组成，类似于一个合唱团由多个歌手以不同的音高签名组成。在地球上可以用NSF的LIGO探测器探测到引力波。这个项目将使用LIGO的数据来确定黑洞发出的引力波是否与爱因斯坦预测的完全一样，或者黑洞是否“唱”出了意想不到的曲调。这样的测试可能会导致物理学的新发现，让我们更好地了解宇宙的基本运作。第二个组成部分是开发一个苹果硅计算机网络，在LIGO数据中搜索新的引力波。这种网络有可能以极低的成本大大加快搜索新信号的速度。这将使资源不足的大学和以本科为重点的学院更容易直接为引力波天文学做出贡献，从而扩大美国基础STEM研究的吸引力和机会。该奖项为学生提供支持，他们将获得广泛适用的国家需要的数据科学技能。该奖项支持开源、基于python的跨维马尔可夫链蒙特卡罗采样器的开发，该采样器将自然地识别由双黑洞合并形成的黑洞发出的一组可观测的准正态模式。这将应用于新的引力波探测。需要解决的关键科学问题包括：在合并时是否可以观察到主导模式的弦外之音，如果可以，是哪些？是否可以观察到其他次主导模式？如果观测到不止一种模式，它们是否符合广义相对论？为了回答这些问题（以及用引力波做任何科学研究），候选信号必须首先被识别出来。目前，这是通过使用数据中心集群上的大量cpu执行匹配过滤器搜索来完成的。由于需要在CPU和GPU之间传输数据，以前利用GPU的努力一直受到阻碍。新的“片上系统”（SoC），如苹果的硅处理器，避开了这个问题，因为内存是在CPU和GPU内核之间共享的。该合同将为SoC处理器集群的构建、软件开发和测试提供资金。其目标是以比当前更快的速度和更低的成本执行搜索。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award supports two projects. The first is to develop methods for testing Einstein's theory of relativity in one of the most extreme environments in the universe: near the horizon of a black hole. Einstein's theory predicts that gravitational waves emitted by black holes should consist of specific frequencies, similar to how a chorus consists of multiple singers signing at different pitches. Gravitational waves are detectable here on Earth with NSF's LIGO detector. This project will use LIGO data to determine if the chorus of gravitational waves emitted by a black hole is exactly as Einstein predicted, or if the black hole "sings" an unexpected tune. Such tests may lead to new discoveries in physics, giving us a better understanding of the fundamental workings of the universe. The second component is to develop a network of Apple Silicon computers to search for new gravitational-waves in LIGO data. Such a network has the potential to make searching for new signals substantially faster and at very low cost. This will make it easier for lesser-resourced universities and undergraduate-focused colleges to directly contribute to gravitational-wave astronomy, broadening the appeal and access to fundamental STEM research in the US. The award provides support for students, who will gain widely-applicable data science skills that are of great national need.This award supports the development of an open-source, Python-based transdimensional Markov-chain Monte Carlo sampler that will naturally identify the set of observable quasi-normal modes emitted by a black hole that is formed in binary black hole mergers. This will be applied to new gravitational-wave detections. Key science questions to be addressed include: are overtones of the dominant mode observable at merger, and if so, which ones? Are other sub-dominant modes observable? If more than one mode is observable, are they consistent with general relativity? In order to answer such questions (and to do any science with gravitational waves), candidate signals must first be identified. Currently this is done by performing a matched-filter search using large numbers of CPUs on data-center clusters. Previous efforts to utilize GPUs have been hampered by the need to transfer data between the CPU and GPU. New "Systems on a Chip" (SoC) such as the Apple Silicon processors side-step this issue, as memory is shared between the CPU and GPU cores. This award will fund the construction, software development, and testing of a cluster of SoC processors. The goal is to perform searches significantly faster and for much lower cost than what is currently done.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.