Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Computing the Wave: Where the Gravitational Wave Community benefits from High-Energy Physics, and where it differs ?

Marco Meyer-Conde, Nobuyuki Kanda, Hirotaka Takahashi, Ken-ichi Oohara, Kazuki Sakai

TL;DR

The paper argues for leveraging cross-domain computing advances between High-Energy Physics and Gravitational Wave science to enhance real-time GW alerting and offline analyses. It presents IGWN's dual-branch computing infrastructure (LLAI for low-latency alerts and Offline Analysis for high-latency processing) and introduces a ROOT+ data-analysis framework that extends ROOT with advanced signal-processing capabilities, including the KFR library and FFT benchmarking against FFTW. The study details prototyping libraries (config parsers, Kafka interfaces, data formats) and demonstrates performance benefits of KFR over FFTW, while integrating LibTorch/ONNX for ML inference within ROOT. OSDF/CVMFS data federation and a modular ROOT+ approach are positioned to bridge HEP and GW software ecosystems, enabling scalable, ML-enabled, real-time and offline analyses for current and future observing runs.

Abstract

High-Energy Physics (HEP) and Gravitational Wave (GW) communities serve different scientific purposes. However, their methodologies might potentially offer mutual enrichment through common software developments. A suite of libraries is currently being prototyped and made available at https://git.ligo.org/kagra/libraries-addons/root, extending at no cost the CERN ROOT data analysis framework toward advanced signal processing. We will also present a performance benchmark comparing the FFTW and KFR library performances.

Computing the Wave: Where the Gravitational Wave Community benefits from High-Energy Physics, and where it differs ?

TL;DR

The paper argues for leveraging cross-domain computing advances between High-Energy Physics and Gravitational Wave science to enhance real-time GW alerting and offline analyses. It presents IGWN's dual-branch computing infrastructure (LLAI for low-latency alerts and Offline Analysis for high-latency processing) and introduces a ROOT+ data-analysis framework that extends ROOT with advanced signal-processing capabilities, including the KFR library and FFT benchmarking against FFTW. The study details prototyping libraries (config parsers, Kafka interfaces, data formats) and demonstrates performance benefits of KFR over FFTW, while integrating LibTorch/ONNX for ML inference within ROOT. OSDF/CVMFS data federation and a modular ROOT+ approach are positioned to bridge HEP and GW software ecosystems, enabling scalable, ML-enabled, real-time and offline analyses for current and future observing runs.

Abstract

High-Energy Physics (HEP) and Gravitational Wave (GW) communities serve different scientific purposes. However, their methodologies might potentially offer mutual enrichment through common software developments. A suite of libraries is currently being prototyped and made available at https://git.ligo.org/kagra/libraries-addons/root, extending at no cost the CERN ROOT data analysis framework toward advanced signal processing. We will also present a performance benchmark comparing the FFTW and KFR library performances.

Paper Structure

This paper contains 6 sections, 6 figures.

Table of Contents

  1. Introduction
  2. IGWN Computing Infrastructure
  3. Low Latency Alert Generation Infrastructure
  4. Offline Analysis Infrastructure
  5. Data Analysis Framework Prototype
  6. Conclusions

Figures (6)

  • Figure 1: Low Latency Alert Generation Infrastructure Workflow Overview.
  • Figure 2: (a) Data volume projections at the time of O3 run; (b) latency estimate from Kamioka mines to Kashiwa ICRR cluster nearby Tokyo around the same time kagra_dmg_ptep.
  • Figure 3: GraceDB Centralized Database Overview highlighting 81 significant detection candidates in O4 as compared to the 90 events in total through O1 to O3
  • Figure 4: Comparison of O3 vs. O4 computing resources in March 2024. igwn_grafana
  • Figure 5: (a) Left: Power Spectrum Density calculation (GW150914 gw150914); (b) Middle: example of digital filter $H(z)$; (c) Right: Matched filtering on c
  • ...and 1 more figures