Parallelized inference for gravitational-wave astronomy
Parallelized inference for gravitational-wave astronomy
复制标题
引力波天文学的并行推理
DOI：
10.1103/physrevd.100.043030
复制
发表时间：
2019
影响因子：
5
通讯作者：
G. Poole
中科院分区：
文献类型：
--
作者：
C. Talbot;Rory J. E. Smith;E. Thrane;G. Poole
贝叶斯推断是引力波天文学的主力，例如，确定合并黑洞的质量和自旋，揭示中子星星状态方程，揭示致密双星的人口特性。这些推论所带来的科学带来了计算成本，这可能会限制我们能够回答的问题。预计这一费用将增加。随着检测器的改进，检测率将上升，允许更少的时间来分析每个事件。低频灵敏度的提高将产生更长的信号，增加每个事件的计算次数。瞬态目录中条目的不断增加将推高人口研究的成本。虽然贝叶斯推理计算不是完全并行的，但关键部分是并行的：计算重力波形和评估似然函数。图形处理器单元（GPU）擅长这种并行计算。我们报告的进展移植引力波推理计算GPU。使用一个代码-如果可用的话，它利用GPU架构-我们比较了使用现代GPU（NVIDIA P100）和CPU（Intel Gold 6140）的计算时间。我们证明了$\ensuremath{\sim} 50\ifmmode\times\else\texttimes\fi {}$的加速用于紧凑的二进制聚结引力波形生成和似然评估，以及超过100\ifmmode\times\else\texttimes\fi {}$用于当前探测器寿命内的人口推断。随着持续发展，可能会进一步改善。我们基于Python的代码是公开的，可以在不熟悉并行计算平台CUDA的情况下使用。
Bayesian inference is the workhorse of gravitational-wave astronomy, for example, determining the mass and spins of merging black holes, revealing the neutron star equation of state, and unveiling the population properties of compact binaries. The science enabled by these inferences comes with a computational cost that can limit the questions we are able to answer. This cost is expected to grow. As detectors improve, the detection rate will go up, allowing less time to analyze each event. Improvement in low-frequency sensitivity will yield longer signals, increasing the number of computations per event. The growing number of entries in the transient catalog will drive up the cost of population studies. While Bayesian inference calculations are not entirely parallelizable, key components are embarrassingly parallel: calculating the gravitational waveform and evaluating the likelihood function. Graphical processor units (GPUs) are adept at such parallel calculations. We report on progress porting gravitational-wave inference calculations to GPUs. Using a single code---which takes advantage of GPU architecture if it is available---we compare computation times using modern GPUs (NVIDIA P100) and CPUs (Intel Gold 6140). We demonstrate speed-ups of $\ensuremath{\sim}50\ifmmode\times\else\texttimes\fi{}$ for compact binary coalescence gravitational waveform generation and likelihood evaluation, and more than $100\ifmmode\times\else\texttimes\fi{}$ for population inference within the lifetime of current detectors. Further improvement is likely with continued development. Our python-based code is publicly available and can be used without familiarity with the parallel computing platform, CUDA.
登录
查看更多内容
DOI：
10.3847/2041-8213/aa9bf6
发表时间：
2017-09
期刊：
The Astrophysical Journal Letters
影响因子：
--
作者：
M. Fishbach;D. Holz
通讯作者：
M. Fishbach;D. Holz
影响因子：
5
作者：
D. Wysocki;R. O’Shaughnessy;Y-L. L. Fang-Y-L.-L.-Fang-90268547;Jacob Lange Center for Computational Relativity;Gravitation;R. I. O. Technology;Computational Science Initiative;Brookhaven National Laboratory
通讯作者：
D. Wysocki;R. O’Shaughnessy;Y-L. L. Fang-Y-L.-L.-Fang-90268547;Jacob Lange Center for Computational Relativity;Gravitation;R. I. O. Technology;Computational Science Initiative;Brookhaven National Laboratory
影响因子：
5
作者：
D. Gerosa;F. H'ebert;L. Stein
通讯作者：
D. Gerosa;F. H'ebert;L. Stein
影响因子：
8.6
作者：
Ajith, P.;Hannam, M.;Seiler, J.
通讯作者：
Seiler, J.
DOI：
10.3847/2041-8213/aad800
发表时间：
2018-05
期刊：
The Astrophysical Journal Letters
影响因子：
--
作者：
M. Fishbach;D. Holz;W. Farr
通讯作者：
M. Fishbach;D. Holz;W. Farr