We introduce an ensemble of artificial intelligence models for gravitational wave detection that we trained in the Summit supercomputer using 32 nodes, equivalent to 192 NVIDIA V100 GPUs, within 2 hours. Once fully trained, we optimized these models for accelerated inference using NVIDIA TensorRT. We deployed our inference-optimized AI ensemble in the ThetaGPU supercomputer at Argonne Leadership Computer Facility to conduct distributed inference. Using the entire ThetaGPU supercomputer, consisting of 20 nodes each of which has 8 NVIDIA A100 Tensor Core GPUs and 2 AMD Rome CPUs, our NVIDIA TensorRT-optimized AI ensemble processed an entire month of advanced LIGO data (including Hanford and Livingston data streams) within 50 seconds. Our inference-optimized AI ensemble retains the same sensitivity of traditional AI models, namely, it identifies all known binary black hole mergers previously identified in this advanced LIGO dataset and reports no misclassifications, while also providing a 3X inference speedup compared to traditional artificial intelligence models. We used time slides to quantify the performance of our AI ensemble to process up to 5 years worth of advanced LIGO data. In this synthetically enhanced dataset, our AI ensemble reports an average of one misclassification for every month of searched advanced LIGO data. We also present the receiver operating characteristic curve of our AI ensemble using this 5 year long advanced LIGO dataset. This approach provides the required tools to conduct accelerated, AI-driven gravitational wave detection at scale.
We optimize the third-generation gravitational-wave detector to maximize the range to detect core-collapse supernovae. Based on three-dimensional simulations for core-collapse and the corresponding gravitational-wave waveform emitted, the corresponding detection range for these waveforms is limited to within our galaxy even in the era of third-generation detectors. The corresponding event rate is two per century. We find from the waveforms that to detect core-collapse supernovae with an event rate of one per year, the gravitational-wave detectors need a strain sensitivity of 3$\times10^{-27}~$…
The Omicron software is a tool developed to perform a multi-resolution time-frequency analysis of data from gravitational-wave detectors: the LIGO, Virgo, and KAGRA detectors. Omicron generates spectrograms from whitened data streams, offering a visual representation of transient detector noises and gravitational-wave events. In addition, these events can be parameterized with an optimized resolution. They can be written to disk to conduct offline noise characterization and gravitational-wave event validation studies. Omicron is optimized to process, in parallel, thousands of data streams reco…
The Advanced LIGO and Virgo gravitational wave observatories have opened a new window with which to study the inspiral and mergers of binary compact objects. These observations are most powerful when coordinated with multi-messenger observations. This was underlined by the first observation of a binary neutron star merger GW170817, coincident with a short Gamma-ray burst, GRB170817A, and the identification of the host galaxy NGC~4993 from the optical counterpart AT~2017gfo. Finding the fast-fading optical counterpart critically depends on the rapid production of a sky-map based on LIGO/Virgo d…
Cyber infrastructure will be a critical consideration in the development of next generation gravitational-wave detectors. The demand for data analysis computing in the 3G era will be driven by the high number of detections as well as the expanded search parameter space for compact astrophysical objects and the subsequent parameter estimation follow-up required to extract the nature of the sources. Additionally, there will be an increased need to develop appropriate and scalable computing cyberinfrastructure, including data access and transfer protocols, and storage and management of software t…
Max-Planck-Institut für Gravitationsphysik（马克斯·普朗克引力物理研究所）；Leibniz Universität Hannover（汉诺威莱布尼茨大学）；University of Rhode Island（罗德岛大学）
Gravitational-wave detection strategies are based on a signal analysis technique known as matched filtering. Despite the success of matched filtering, due to its computational cost, there has been recent interest in developing deep convolutional neural networks (CNNs) for signal detection. Designing these networks remains a challenge as most procedures adopt a trial and error strategy to set the hyperparameter values. We propose a new method for hyperparameter optimization based on genetic algorithms (GAs). We compare six different GA variants and explore different choices for the GA-optimized…
Results of the deepest all-sky survey for continuous gravitational waves on LIGO S6 data running on the Einstein@Home volunteer distributed computing project
Two new methods have been proposed for solving the gravitational constraints without using elliptic solvers by formulating them as either an algebraic-hyperbolic or parabolic-hyperbolic system. Here, we compare these two methods and present a unified computational infrastructure for their implementation as numerical evolution codes. An important potential application of these methods is the prescription of initial data for the simulation of black holes. This paper is meant to support progress and activity in that direction.
The detection of intermediate-mass black holes (IMBHs) i.e. those with mass $\sim 100$-$10^5 M_\odot$, is an emerging goal of gravitational-wave (GW) astronomy with wide implications for cosmology and tests of strong-field gravity. Current PyCBC-based searches for compact binary mergers, which matched filter the detector data against a set of template waveforms, have so far detected or confirmed several GW events. However, the sensitivity of these searches to signals arising from mergers of IMBH binaries is not optimal. Here, we present a new optimised PyCBC-based search for such signals. Our …
Automated Algorithmic Discovery for Scientific Computing through LLM-Guided Evolutionary Search: A Case Study in Gravitational-Wave Detection
International Centre for Theoretical Physics Asia-Pacific（国际理论物理中心亚太中心）；University of Chinese Academy of Sciences（中国科学院大学）；Taiji Laboratory for Gravitational Wave Universe（引力波宇宙Taiji实验室）；Tsinghua University（清华大学）
Gravitational waves have been detected from the inspiral of a binary neutron-star, GW170817, which allowed constraints to be placed on the neutron star equation of state. The equation of state can be further constrained if gravitational waves from a post-merger remnant are detected. Post-merger waveforms are currently generated by numerical-relativity simulations, which are computationally expensive. Here we introduce a hierarchical model trained on numerical-relativity simulations, which can generate reliable post-merger spectra in a fraction of a second. Our spectra have mean fitting factors…
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）
We use artificial intelligence (AI) to learn and infer the physics of higher order gravitational wave modes of quasi-circular, spinning, non precessing binary black hole mergers. We trained AI models using 14 million waveforms, produced with the surrogate model NRHybSur3dq8, that include modes up to $\ell \leq 4$ and $(5,5)$, except for $(4,0)$ and $(4,1)$, that describe binaries with mass-ratios $q\leq8$, individual spins $s^z_{\{1,2\}}\in[-0.8, 0.8]$, and inclination angle $θ\in[0,π]$.Our probabilistic AI surrogates can accurately constrain the mass-ratio, individual spins, effective spin, a…
labrador: A domain-optimized machine-learning tool for gravitational wave inference
TAPIR, California Institute of Technology（加州理工学院 TAPIR）；Kavli Institute for Cosmological Physics, The University of Chicago（芝加哥大学卡弗里宇宙物理学研究所）；School of Natural Sciences, Institute for Advanced Study（高等研究院自然科学学院）；Dipartimento di Fisica, Università di Pisa（比萨大学物理系）；LIGO Laboratory, California Institute of Technology（加州理工学院 LIGO 实验室）
We present LEGWORK (LISA Evolution and Gravitational Wave Orbit Kit), an open-source Python package for making predictions about stellar-origin gravitational wave sources and their detectability in LISA or other space-based gravitational wave detectors. LEGWORK can be used to evolve the orbits of sources due to gravitational wave emission, calculate gravitational wave strains (using post-Newtonian approximations), compute signal-to-noise ratios and visualise the results. It can be applied to a variety of potential sources, including binaries consisting of white dwarfs, neutron stars and black …
Parameter-space metric for all-sky semicoherent searches for gravitational-wave pulsars
全天空搜索引力波星的灵敏度主要受到计算资源有限的限制。半相干搜索是一种以固定计算成本最大化对引力波探测器灵敏度的广泛使用的方法：将来自引力波探测器的数据划分为多个段，对每个段进行相干分析，并将每个段的分析结果相加在一起。生成模板银行的相干分析的每一个部分，并为求和，需要的知识的度量相关的相干和半相干的参数空间分别。我们提出了一个有用的近似半相干参数空间度量，类似于Wette和Prix [Phys. Rev. D 88，123005（2013）]中提出的相干度量。新的半相干度量与Pletsch [Phys. Rev. D 82，042002（2010）]和Brady和Creighton [Phys. Rev. D 61，082001（2000）]中的先前工作进行了比较。我们发现，半相干全天空搜索需要的数量级比以前预测的模板。
In our previous research, simulation showed that a quantum locking scheme with homodyne detection in sub-cavities is effective in surpassing the quantum noise limit for Deci-hertz Interferometer Gravitational Wave Observatory (DECIGO) in a limited frequency range. This time we have simulated an optical spring effect in the sub-cavities of the quantum locking scheme. We found that the optimized total quantum noise is reduced in a broader frequency band, compared to the case without the optical spring effect significantly improving the sensitivity of DECIGO to the primordial gravitational waves.
The search for continuous gravitational waves from unknown isolated sources is computationally limited due to the enormous parameter space that needs to be covered and the weakness of the expected signals. Therefore semi-coherent search strategies have been developed and applied in distributed computing environments such as Einstein@Home, in order to narrow down the parameter space and identify interesting candidates. However, in order to optimally confirm or dismiss a candidate as a possible gravitational-wave signal, a fully-coherent follow-up using all the available data is required. We pre…
We study post-inflationary evolution in $α$-attractor T-models of inflation. We consider the dynamics of both scalar fields present in these models: the inflaton and the spectator, as a negative field-space curvature may lead to geometrical destabilization of the spectator. We perform state-of-the-art lattice simulations with a dedicated numerical code optimized for those models. We corroborate earlier findings that the perturbations of the spectator field are much more unstable than the perturbations of the inflaton field, so the dynamics of early stages of preheating is dominated by the evol…
