a Department of Physics, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, USA
b Data Science and Learning Division, Argonne National Laboratory, Lemont, Illinois 60439, USA
c National Center for Supercomputing Applications, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, USA
d Department of Computer Science, University of Chicago, Chicago, Illinois 60637, USA
e Leadership Computing Facility, Argonne National Laboratory, Lemont, Illinois 60439, USA
arXiv: arxiv/2110.06968
We present a deep-learning artificial intelligence model that is capable of learning and forecasting the late-inspiral, merger and ringdown of numerical relativity waveforms that describe quasi-circular, spinning, non-precessing binary black hole mergers. We used the NRHybSur3dq8 surrogate model to produce train, validation and test sets of \(\ell=|m|=2\) waveforms that cover the parameter space of binary black hole mergers with mass-ratios \(q\leq8\) and individual spins \(|{s^z_{\{1,\,2\}}}| \leq 0.8\). These waveforms cover the time range \(t\in[-5000\textrm{M}, 130\textrm{M}]\), where \(t=0M\) marks the merger event, defined as the maximum value of the waveform amplitude. We harnessed the ThetaGPU supercomputer at the Argonne Leadership Computing Facility to train our AI model using a training set of 1.5 million waveforms. We used 16 NVIDIA DGX A100 nodes, each consisting of 8 NVIDIA A100 Tensor Core GPUs and 2 AMD Rome CPUs, to fully train our model within 3.5 hours. Our findings show that artificial intelligence can accurately forecast the dynamical evolution of numerical relativity waveforms in the time range \(t\in[-100\textrm{M}, 130\textrm{M}]\). Sampling a test set of 190,000 waveforms, we find that the average overlap between target and predicted waveforms is \(\gtrsim 99\%\) over the entire parameter space under consideration. We also combined scientific visualization and accelerated computing to identify what components of our model take in knowledge from the early and late-time waveform evolution to accurately forecast the latter part of numerical relativity waveforms. This work aims to accelerate the creation of scalable, computationally efficient and interpretable artificial intelligence models for gravitational wave astrophysics.
Please refer to our paper for a full discussion of Methods and Results.
Here we provide the interactive version of a select number of figures presented in the paper. You can zoom, hover over plots, and interact with the legend.
These subplots correspond to Fig. 7 in the paper and still show only one of the 12 Attention heads,
but for several different input waveforms.
Left panel Heatmap for one of the
twelve \(\textit{cross attention}\) heads showing which parts of the input waveform (y-axis) the decoder
is paying attention to when predicting the output
at any particular timestep (x-axis).
Right panel Heatmap showing one of the
\(\textit{self attention}\) heads of the decoder.
These are additional visualizations of cross- and self-attention heads. t_in corresponds to
the time-steps of the input waveform and t_out corresponds to the time-steps of the predicted waveform.
Select a portion of the t_out axis and slide it up and down to see how the model pays "attention" to
different segments of the input waveform when predicting the selected portion of t_out.
Please cite this webpage as
@Misc{KHAN2110.06968_interactive_results,
author = "Asad Khan",
title = "Interpretable AI forecasting for numerical relativity waveforms of quasi-circular, spinning, non-precessing binary black hole mergers",
year = "2021",
url = "https://khanx169.github.io/gw_forecasting/interactive_results.html"
}