BIMSA >
相对论物理讨论班
相对论物理讨论班
From Signals to Sources: Machine Learning for Space-Borne Gravitational-Wave Astronomy
From Signals to Sources: Machine Learning for Space-Borne Gravitational-Wave Astronomy
组织者
演讲者
Xue-Ting Zhang
时间
2026年06月17日 14:30 至 15:30
地点
A3-2-301
线上
Zoom 928 682 9093
(BIMSA)
摘要
Future space-borne gravitational-wave observatories like TianQin will observe compact binaries for months to years before merger, offering unprecedented opportunities for gravitational-wave and multi-messenger astronomy. However, the long-duration signals and high-dimensional parameter space pose significant challenges for traditional data-analysis techniques. Template-bank searches become increasingly expensive, while Bayesian parameter estimation often requires computationally intensive sampling methods. This talk presents a machine-learning framework for accelerating gravitational-wave data analysis in the millihertz band. After reviewing the main computational bottlenecks in classical search and inference pipelines, the talk introduces convolutional neural networks for signal detection and parameter point estimation of stellar-mass binary black holes, followed by a neural posterior estimation method based on neural spline flows for rapid Bayesian inference and pre-merger localization of massive black hole binaries. Together, these studies illustrate how machine learning can support the transition from signal detection to source characterization, enabling low-latency science with future space-borne gravitational-wave detectors.
演讲者介绍
Dr. Xue-Ting Zhang is a Humboldt Research Fellow at the Max Planck Institute for Gravitational Physics (Albert Einstein Institute). Her research focuses on gravitational-wave and multi-messenger astronomy, especially data-analysis methods for future space-borne detectors. Her work includes long-duration gravitational-wave signal searches, and fast inference pipelines designed to support electromagnetic follow-up. Through studies of extreme mass-ratio inspirals, stellar-mass binary black-hole early inspirals, and massive black-hole binary, her research connects advanced data-driven methods with the scientific goals of space-based gravitational-wave astronomy.