EVENTO
New Markov chain Monte Carlo methods applied to high-dimensional stochastic inverse problems in subsurface flows
Tipo de evento: Seminário de Avaliação - Série A
Uncertainty quantification for the solution of inverse problems for subsurface applications often relies on Markov chain Monte Carlo (McMC) methods, but high stochastic dimensionality remains a key challenge, especially in reservoir simulation. While dimensional reduction techniques such as the KarhunenLoève Expansion (KLE) and Variational Autoencoders (VAE) provide more compact representations of random inputs, high-constrast permeability fields along with larger computational domains still lead to high-dimensional, challenging inverse problems. In this context, efficient sampling strategies are essential to accelerate parameter space exploration; building on prior work from the qualification exam, this thesis introduces the Preconditioned CrankNicolson Differential Evolution MCMC (pCN-DE) method, which combines the variance control of pCN with the exploratory efficiency of DE-MCMC while requiring significantly fewer chains (about half the stochastic dimension, rather than the usual double or more). This seminar presents preliminary results of pCN-DE applied to a 2D stationary single-phase flow problem and initial studies on pollutant transport, along with an outline of the research plan for the remainder of the doctoral work.Evento HíbridoLocal: Auditório LNCCLink da sala virtual: meet.google.com/mfv-xpeh-yat
Data Início: 08/06/2026 Hora: 14:00 Data Fim: 08/06/2026 Hora: 17:00
Local: LNCC - Laboratório Nacional de Computação Ciêntifica - Auditorio B
Aluno: Michel Antonio Tosin Caldas - - LNCC
Orientador: Luis Felipe Feres Pereira - - Marcio Rentes Borges - Laboratório Nacional de Computação Científica - LNCC
Participante Banca Examinadora: Helio Pedro Amaral Souto - Universidade do Estado do Rio de Janeiro - UERJ José Karam Filho - Laboratório Nacional de Computação Científica - LNCC Marcio Rentes Borges - Laboratório Nacional de Computação Científica - LNCC
Suplente Banca Examinadora: Fabio Lima Custodio - Laboratório Nacional de Computação Científica - LNCC Renato Simões Silva - Laboratório Nacional de Computação Científica - LNCC
