Gravitational-wave Inference with Waveform Uncertainty

Project Title: Gravitational-wave Inference with Waveform Uncertainty
Program:
CAREERS (323)

Project Image: https://support.access-ci.org/system/files/webform/project/3620/original.jpg
Tags:
astrophysics (297), benchmarking (400), deep-learning (303), gravitational-waves (597), machine-learning (272)

Status: Complete
Project Leader
--------------
Project Leader:
Michael Puerrer

Email: mpuerrer@uri.edu
Mobile Phone: {Empty}
Work Phone: {Empty}

Project Personnel
-----------------
Mentor(s):
Gaurav Khanna (541)

Student-facilitator(s):
Ritesh Bachhar (1790)

Mentee(s):
{Empty}


Project Information
-------------------
Project Description:
Models of gravitational waves (GW) emitted from merging black hole binaries are based on numerical data and modeling assumptions which introduces sources of error. When performing Bayesian inference of the binary parameters from a GW signal these models are used to compute the posterior distribution with a sampling algorithm. This project aims at including marginalization over waveform uncertainty at the inference stage. The resulting posterior distribution is expected to peak closer to the true signal parameters at the cost of a slight broadening of the distribution, trading precision for improved accuracy.

In this project the student will adapt computational inference workflows to execute on URI’s UNITY cluster and perform extensive testing using real and synthetic GW events while marginalizing over internal degrees of freedom of an effective-one-body waveform model. If time permits, the student will use the developed workflow to compare two methodologies: (i) a probabilistic Gaussian process regression (GPR) waveform model using a standard stochastic sampler, and (ii) using the novel technique of neural posterior estimation on training set data augmented with waveform uncertainties, leveraging the deep learning-based Dingo inference code.


Project Information Subsection
------------------------------
Project Deliverables:
{Empty}

Project Deliverables:
{Empty}

Student Research Computing Facilitator Profile:
{Empty}

Mentee Research Computing Profile:
{Empty}

Student Facilitator Programming Skill Level: {Empty}
Mentee Programming Skill Level: {Empty}
Project Institution: University of Rhode Island -- Center for Computational Research
Project Address:
{Empty}

Anchor Institution: CR-University of Rhode Island
Preferred Start Date: {Empty}
Start as soon as possible.: No
Project Urgency: Already behind3Start date is flexible
Expected Project Duration (in months): 3
Launch Presentation: https://support.access-ci.org/system/files/webform/project/3620/wf_uncertainty_launch.pptx
Launch Presentation Date: 06/14/2023
Wrap Presentation: https://support.access-ci.org/system/files/webform/project/3620/wf_uncertainty_wrap.pdf
Wrap Presentation Date: 08/16/2023
Project Milestones:
- Milestone Title: Milestone #1
  Milestone Description: Student reviews background materials and familiarizes themself with the key computational Python packages https://git.ligo.org/lscsoft/bilby and https://github.com/dingo-gw/dingo; launch presentation; code repository. 
  Completion Date Goal: 2023-06-30
- Milestone Title: Milestone #2
  Milestone Description: Student sets up the Bilby code on UNITY and performs a series of analyses on synthetic GW signals at different signal-to-noise ratios with the GPR model, along with standard models for comparisons.
  Completion Date Goal: 2023-07-31
- Milestone Title: Milestone #3
  Milestone Description: Student builds a training set from GPR model waveforms and trains a neural network using the Dingo code, as well as networks for standard waveform models, for comparison; this is then used for inference on GW signals as in milestone #2, including on real GW events (Training a single network takes about a week on one NVIDIA A100 GPU, while training set construction takes several hours on a single CPU node using dozens of threads. Amortized inference on one NVIDIA A100 GPU then takes less than a minute per GW signal.); wrap presentation; repo updated.
  
  Completion Date Goal: 2023-08-31

Github Contributions: {Empty}
Planned Portal Contributions (if any):
{Empty}

Planned Publications (if any):
{Empty}

What will the student learn?:
{Empty}

What will the mentee learn?:
{Empty}

What will the Cyberteam program learn from this project?:
{Empty}

HPC resources needed to complete this project?:
{Empty}

Notes:
{Empty}



Final Report
------------
What is the impact on the development of the principal discipline(s) of the project?:
{Empty}

What is the impact on other disciplines?:
{Empty}

Is there an impact physical resources that form infrastructure?:
{Empty}

Is there an impact on the development of human resources for research computing?:
{Empty}

Is there an impact on institutional resources that form infrastructure?:
{Empty}

Is there an impact on information resources that form infrastructure?:
{Empty}

Is there an impact on technology transfer?:
{Empty}

Is there an impact on society beyond science and technology?:
{Empty}

Lessons Learned:
{Empty}

Overall results:
{Empty}