Nomination for Early Career Co-Chair

hewang@ucas.ac.cn | 0000-0002-1353-391X | iphysresearch.github.io/blog

ICTP-AP & UCAS
LISA Consortium Core Member
LVK MLA Group Co-Chair

Research focus: GW data analysis, simulation-based inference,  scalable ML methods

LISA Data Analysis R&D Working Group

Current Activities & Background

Consortium & coordination activities

• Co-Chair, LVK Machine Learning Algorithms Group
• Core member, LISA Consortium
• UCAS joined LVK as member institution in 2024
• Current DataAnalysisRDWG project:

  • datard-wg-2025-03 — Novel methods for LISA

Research directions

• simulation-based inference
• probabilistic modeling
• global-fit methodologies
• interpretable ML approaches
• scalable GW data-analysis pipelines

github.com/iphysresearch/GWData-Bootcamp​

Taiji Data Challenge (TDC) tutorial

    GWData-Bootcamp

• Numerical orbits (of Taiji)
• Unequal-arm
• TDI-2.0

github.com/TriangleDataCenter/Triangle-Simulator

Training & community evidence 

• GWData‑Bootcamp: Structured hands‑on training from Python to DL for GWs
• Co‑organized TDCII training sessions
• Active community engagement across LISA / Taiji / TianQin

Methodological Work Relevant to the WG

Current methodological interests

• benchmark datasets/reproducibility
• alternative global-fit strategies
• interpretable AI methodologies
• RL-assisted planning strategies

Bridging methodological research and practical analysis infrastructure

What I Think the WG Needs Most

Shared benchmark & infrastructure development

• realistic validation datasets
• reusable workflows
• common evaluation metrics
• interfaces to DDPC / analysis pipelines

Cross-ecosystem coordination experience

• shared methodological challenges across LISA/Taiji/TianQin
• opportunities for open collaboration and knowledge exchange
• shared tutorials, datasets, and benchmarking activities
• encouraging open-source and reusable analysis components

Lowering the barrier for ML & GW data analysis

• many groups are exploring AI methods for overlap/global-fit problems
• practical training resources remain limited
• need for tutorials on:
  • GW data analysis basics
  • SBI/ML workflows
  • reproducible benchmarking

Improving technical communication

• methodological exchange across projects remains fragmented
• useful developments are often difficult to track or reproduce
• regular technical sharing could help:
  • tutorials / hackathons / focused workshops / benchmark reports

Space-GW analysis is an ecosystem problem

Image credit: TDCII&MH Du

• Part One: Programming Development Environment and Workflow
  • - Linux Commands and Shell Scripting
    - Git Version Control (GitHub / GitLab)
    - SSH Remote Server Access (Shell / VSCode)
    - Containerization with Docker
    - Hands-On: Setting up Python / Jupyter Development Environment
    - Hands-On: Compiling LALsuite / LISAcode Source Code
• Part Two: Python-Based Data Analysis Fundamentals
  • - Introduction to Python Programming
    - Algorithms with Numpy / Pandas / Scipy
    - Hands-On: Exploratory Data Analysis of GW Event Catalog / Glitch Data
    - Hands-On: Matched Filtering for GW150914 Data
    - Data Visualization in Python: Theory and Practice
    - Hands-On: Reproducing Figures from GWTC Papers
• Part Three: Basics of Machine Learning
  • - Overview of Artificial Intelligence
    - Definitions, Objectives, and Types of Machine Learning
    - Machine Learning Project Development and Preparation
    - Hands-On: Clustering Analysis of LIGO's Glitch Data
• Part Four: Introduction to Deep Learning
  • - Overview of Deep Learning Technologies
    - Fundamentals of Artificial Neural Networks (ANN)
    - Convolutional Neural Networks (CNN)
    - Hands-On: Identifying Gravitational Waves from Binary Black Hole Systems using CNN
    - Frontiers of Gravitational Wave Data Analysis and AI

Outline from GWData-Bootcamp: github.com/iphysresearch/GWData-Bootcamp​

Why Coordination Matters Now

Rapid diversification of methodological approaches

• ongoing global-fit activities
• ML/SBI approaches
• alternative representations
• probabilistic inference
• search/planning methods

The challenge is no longer only developing methods

• connecting methods to realistic datasets
• validating across workflows
• improving reproducibility
• reducing duplicated effort
• enabling broader participation

Potential role of the WG

• facilitating technical exchange
• maintaining methodological visibility
• encouraging shared validation efforts
• helping emerging methods mature into reusable infrastructure

Hierarchical Bayesian population inference for LISA (2026)

arXiv: 2601.04168

RL for global-fit (2026)

HW+, arXiv: 26xx.xxxx