580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can a posteriori error estimation techniques be developed and validated for semilinear parabolic partial differential equations using finite element methods?
This research theme focuses on deriving rigorous and computable a posteriori error bounds for semilinear parabolic PDEs discretized by finite element methods, with particular attention to semidiscrete spatial discretization, combination of discontinuous Galerkin (dG) in time and continuous Galerkin (cG) in space methods, and handling nonlinear reaction terms with Lipschitz or local Lipschitz growth conditions. This is critical for enabling adaptive mesh refinements and time stepping to improve computational efficiency and accuracy in solving time-dependent PDEs.
Key finding: Established optimal order a posteriori error bounds in L∞(L2) norm for semidiscrete conforming finite element approximations of semilinear parabolic problems using elliptic reconstruction and energy techniques. The results... Read more
Key finding: Derived a posteriori error bounds for semilinear parabolic equations discretized by discontinuous Galerkin time-stepping combined with continuous finite elements in space, utilizing time reconstruction techniques and elliptic... Read more
Key finding: Provided a comprehensive treatment of space–time dG finite element methods for nonlinear parabolic PDEs, including a priori and a posteriori energy-norm error bounds under only locally Lipschitz nonlinearities without... Read more
2. What are the mathematical frameworks and computational strategies for obtaining robust a posteriori error estimators in mixed finite element approximations of nearly incompressible elasticity problems?
This theme investigates the derivation and validation of a posteriori error estimators that maintain robustness regardless of the Lamé parameter degeneracy as the Poisson ratio approaches 0.5, which causes locking in elasticity simulations. Mixed finite element methods adopting auxiliary pressure variables (Herrmann formulation) are examined, alongside error indicators and bounds that remain stable in the incompressible limit, providing foundations for efficient adaptive mesh refinement in elasticity computations.
Key finding: Developed and analysed several novel a posteriori error estimators for mixed finite element methods of nearly incompressible elasticity and proved that their energy-norm reliability and efficiency bounds do not depend on the... Read more
Key finding: This study numerically validates a probabilistic framework to compare accuracy between finite element spaces P_k and P_m, especially highlighting cases where lower degree elements outperform higher degree ones—a phenomena... Read more
3. How can error estimation frameworks account for finite precision, computational rounding, and model inaccuracies in iterative numerical methods and simulation workflows?
This theme addresses the realistic sources of error beyond discretization, notably rounding errors arising from floating-point arithmetic and approximate computations, as well as model discrepancies. It covers methodologies for error quantification in iterative solvers such as the Conjugate Gradient method in finite precision, encapsulated error estimation approaches for floating-point computations, and statistical or Bayesian frameworks to handle model error and propagate uncertainty in simulations, essential for ensuring reliability and informed decision-making in scientific computing.
Key finding: Provided a detailed analysis of A-norm error lower bounds in the conjugate gradient method, bridging theoretical exact arithmetic error estimates with observed finite precision behavior. The work elucidates why certain error... Read more
Key finding: Introduced encapsulated error arithmetic that simultaneously tracks computed results and associated numerical errors at every computational step, providing a direct, low-overhead mechanism competitive with state-of-the-art... Read more
Key finding: Developed Bayesian statistical methods to quantify and incorporate model error explicitly in predictions, particularly for space-time processes. The framework treats model outputs as biased observations and employs stochastic... Read more