Deep Learning for Gravitational Wave Detection and Analysis

— by He Wang (ICTP-AP, UCAS)

1. Littenberg, Tyson B., and Neil J. Cornish. “Prototype Global Analysis of LISA Data with Multiple Source Types.” arXiv, January 9, 2023. http://arxiv.org/abs/2301.03673.
2. Strub S H, Ferraioli L, Schmelzbach C, et al. Global analysis of LISA data with Galactic binaries and massive black hole binaries[J]. Physical Review D, 2024, 110(2): 024005.
3. Strub S H. Advancing Global Parameter Estimation of Gravitational Waves Recorded by LISA[D]. ETH Zurich, 2024.
4. Bandopadhyay D, Moore C J. GPU-accelerated semicoherent hierarchical search for stellar-mass binary inspiral signals in LISA[J]. Physical Review D, 2024, 110(10): 103026.
5. Houba N, Strub S H, Ferraioli L, et al. Detection and prediction of future massive black hole mergers with machine learning and truncated waveforms[J]. Physical Review D, 2024, 110(6): 062003.
6. Hoy C, Weaving C R, Nuttall L, et al. A rapid multi-modal parameter estimation technique for LISA[J]. Classical and Quantum Gravity, 2024.
7. Alvey J, Bhardwaj U, Domcke V, et al. Leveraging Time-Dependent Instrumental Noise for LISA SGWB Analysis[J]. arXiv preprint arXiv:2408.00832, 2024.  https://github.com/Mauropieroni/GW_response

8. Karnesis, Nikolaos, Michael L Katz, Natalia Korsakova, Jonathan R Gair, and Nikolaos Stergioulas. “Eryn: A Multipurpose Sampler for Bayesian Inference.” Monthly Notices of the Royal Astronomical Society 526, no. 4 (December 21, 2023): 4814–30. https://doi.org/10.1093/mnras/stad2939.

9. Lackeos K, Littenberg T B, Cornish N J, et al. The LISA Data Challenge Radler analysis and time-dependent ultra-compact binary catalogues[J]. Astronomy & Astrophysics, 2023, 678: A123.

15 minutes (10+5)

• Space-based gravitational wave program and scientific objectives
• What is global fitting, and why is it important?
  • The bridge to science
• The Data Analysis Challenge
• Implementing the Global Fit
• Current Global Fit Implementations
  • 1,2,3,4...
• Future Plans
• Main takeaways

GW Astronomy

Gravitational waves (GW) are a strong field effect in the GR.

GW detection opens a new cosmic window

GW Astronomy

Fundamental Physics

• Existence of gravitational waves
• To put constraints on the properties of gravitons

Astrophysics

• Refine our understanding of stellar evolution
• and the behavior of matter under extreme conditions.

Cosmology

• The measurement of the Hubble constant
• Dark energy

Exploring Beyond General Relativity

• Much of the discussion on model generalization has been within the GR framework.
• Our work on beyond General Relativity (bGR) aims to demonstrate AI's potential advantages in detecting signals that surpass GR's limitations.

B. P. Abbott et al. (LIGO-Virgo), PRD 100, 104036 (2019). 

Yu-Xin Wang, Xiaotong Wei, Chun-Yue Li, Tian-Yang Sun, Shang-Jie Jin, He Wang*, Jing-Lei Cui, Jing-Fei Zhang, and Xin Zhang*. arXiv:2410.20129

AI for Science

• 2024.5: Science in the age of AI: How artificial intelligence is changing the nature and method of scientific research (Royal Soc.)
• 2024.4：美国总统科学技术顾问委员会（PCAST）发布《赋能研究：利用人工智能应对全球挑战》报告
• 2023.3：为贯彻落实国家《新一代人工智能发展规划》，科技部会同自然科学基金委近期启动“人工智能驱动的科学研究”专项部署工作。

AI for Science

The core driving force of AI4Sci largely lies in its “interpolation” generalization capabilities, showcasing its powerful complex modeling abilities.

From 李宏毅

AI for Science

The core driving force of AI4Sci largely lies in its “interpolation” generalization capabilities, showcasing its powerful complex modeling abilities.

AI for Science

The core driving force of AI4Sci largely lies in its “interpolation” generalization capabilities, showcasing its powerful complex modeling abilities.

2103.01641

2407.07229

Parameter Inference

• Calculating likelihood function
• Template generation time-consuming

Bayesian statistics

SBI methods

• Simulation-Based Inference (SBI), also known as likelihood-free inference (LFI) or implicit likelihood inference (ILI).

Bo Liang, He Wang*, In preparation.

Bo Liang, He Wang*, In preparation.

PRL 127, 24 (2021) 241103.

PRL 130, 17 (2023) 171403.

...

DINGO

Fast Parameter Inference on Pulsar Timing Arrays with Normalizing Flows

PRL 133, 011402 (2024)

PRD 109, 12354 (2024)

Neural Posterior Estimation with Guaranteed Exact Coverage: The Ringdown of GW150914

PRD 108, 4 (2023): 044029.

Bo Liang, Minghui Du*, He Wang*, et al.  MLST 5,  no. 4 (2024): 045040

Space-borne​
GW Detector

LISA / TaiJi / TianQin program

The Bridge to Science​

Disentangling All Sources:

Fit all parameters of all astrophysical signals and instrumental features observed simultaneously and comprehensively.

1. Galactic Binaries

• Study the formation and evolution of compact binary stars and the structure of the Milky Way Galaxy

The analysis of the best currently known LISA binaries, even making maximal use of the available information about the sources, is susceptible to ambiguity or biases when not simultaneously fitting to the rest of the galactic population.          (copied from Littenberg et al. 2404.03046)

credit: Karnesis et al, arXiv:2303.02164

credit: Kupfer et al, arXiv:2302.12719

credit: Karnesis et al, arXiv:2303.02164v2

2. Massive Black Holes

• Trace the origins, growth and merger histories of massive Black Holes across cosmic epochs

The addition of GBs biases the parameter recovery of masses and spins away from the injected values, reinforcing the need for a global fit pipeline which will simultaneously fit the parameters of the GB signals before estimating the parameters of MBHBs.  
                          (Copied from Weaving, et al. CQG, 2023, 41(2): 025006.)

3. Extreme and Intermediate Mass Ratio Inspirals

• Probe the properties and immediate environments of Black Holes in the local Universe using extreme mass-ratio inspirals and intermediate mass-ratio inspirals

4. Stellar Mass Black Holes

• Understand the astrophysics of stellar-mass Black Holes

LISA Data Challenge 1b: Yorsh
https://lisa-ldc.lal.in2p3.fr/challenge1b

5. Fundamental Nature of Gravity

• Explore the fundamental nature of gravity and Black Holes

6. Expansion of the Universe

7. Stochastic Gravitational Wave Backgrounds

• Probe the rate of expansion of the Universe with standard sirens

• Understand stochastic gravitational wave backgrounds and their implications for the early Universe and TeV-scale particle physics
   
• Search for gravitational wave bursts and unforeseen sources

Current
Global Fit Implementations

Once all is removed that can be removed, that is how designs are truly in their simplest form.

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

（Statistics as of December 2024, covering only MBHB and UCB-related. If there are any discrepancies, please let me know.)

• Tempered RJMCMC
• Information build up over time

• Maximum Likelihood Estimate
• Week by week build up
• GPU!

Credit: 杜明辉

• Density fit for posterior distributions

  • use the old posterior to form a proposal for the extended data.

• Density fit for the Galaxy
  • fitt a Galaxy model for joint distribution for \((A, \beta, \lambda)\).
• ...

Global-fit: AI proposal

• Density fit for posterior distributions

  • use the old posterior to form a proposal for the extended data.

• Density fit for the Galaxy
  • fitt a Galaxy model for joint distribution for \((A, \beta, \lambda)\).
• ...

Global-fit: AI proposal

• Ashton, G, and C Talbot. MNRAS 507, no. 2 (2021): 2037–51.
• Korsakova, N, et al. (2402.13701)
• Wouters, T, et al. (2404.11397​)

Main takeaways

• Multiple pipelines and their outputs available
• Still preliminary; much remains to be done
• Additional software engineering support and a comprehensive mock data challenge/benchmark are needed.

Future Plans

• Independently developed and implemented
• Python support
• GPU driven
• AI powered

THANK YOU

If you have any questions, feel free to ask now or email me: 

      hewang@ucas.ac.cn