Use of Conditional Variational Autoencoders for parameter estimation of microlensed gravitational waves

Gravitational waves (GWs) are ripples in the fabric of spacetime, first predicted by Albert Einstein in 1916 as a consequence of his general theory of relativity. These waves are generated by the acceleration of massive objects, such as the collision of black holes or neutron stars, and propagate ou...

Descripción completa

Detalles Bibliográficos
Autor: Bada Nerín, Roberto
Tipo de recurso: tesis de maestría
Fecha de publicación:2025
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/428985
Acceso en línea:https://hdl.handle.net/2117/428985
Access Level:acceso abierto
Palabra clave:Gravitational waves
Deep learning (Machine learning)
Black holes (Astronomy)
Ones gravitacionals
autoencoder
variational autoencoder
conditional variational autoencoder
gravitació
relativitat general
lensing
microlensing
estimació de paràmetres
Ligo
Virgo
Kagra
col·laboració LVK.
gravitation
general relativity
parameter estimation
LVK collaboration
Aprenentatge profund
Forats negres (Astronomia)
Àrees temàtiques de la UPC::Informàtica::Intel·ligència artificial::Aprenentatge automàtic
Descripción
Sumario:Gravitational waves (GWs) are ripples in the fabric of spacetime, first predicted by Albert Einstein in 1916 as a consequence of his general theory of relativity. These waves are generated by the acceleration of massive objects, such as the collision of black holes or neutron stars, and propagate outward at the speed of light. By analyzing the frequency and amplitude of GW signals, scientists can study the dynamics of compact binary systems, measure the properties of black holes, and explore the extreme physics of high-energy astrophysical events. As GWs traverse the cosmos, their signals can be influenced by massive objects along their path, leading to a phenomenon known as gravitational lensing. This effect, analogous to the lensing of light, occurs when the gravitational f ield of an intervening mass bends and magnifies the passing wave, providing a unique opportunity to study cosmology and astrophysics at multiple scales. Detecting microlensing signatures–i.e. GWs lensed by relatively small masses–, in particular, requires efficient parameter estimation methods due to the high computational cost of traditional Bayesian inference. In this work we explore the use of deep learning, namely Conditional Variational Autoencoders (CVAE), to estimate parameters of microlensed binary black hole (simulated) waveforms. We find that our CVAE model yields accurate parameter estimation and significant computational savings compared to Bayesian methods such as bilby (up to five orders of magnitude faster inferences). Moreover, the incorporation of CVAE-generated priors in bilby reduces the average runtime of the latter in about 48% with no penalty on its accuracy. Our results suggest that a CVAE model is a promising tool for future low-latency searches of lensed signals. Further applications to actual signals and integration with advanced pipelines could help extend the capabilities of GW observatories in detecting microlensing events.