Core Concepts
Full space-time numerical solution of advection-diffusion equation using Galerkin/least-square method ensures stability and accuracy.
Abstract
The content discusses the application of a continuous Galerkin finite element method to solve the advection-diffusion equation in a full space-time formulation. It explores stability, computational cost, error estimates, and adaptive space-time mesh refinement. The article presents examples illustrating convergence, stability analysis, and comparison with conventional time-marching algorithms.
Introduction
Dual discretization in transient problems.
Space-time formulation for improved parallel performance.
Alternative Strategy
Formulating problems in space-time domain.
Benefits of parallelism in both space and time.
Historical Context
Early references to space-time formulations.
Recent interest due to computational resources.
Stability Analysis
Stability concerns in parabolic equations through space-time methods.
Use of Galerkin/least squares approach for stability.
Computational Cost Reduction
Increased computational cost with additional dimensions.
Efficiency through adaptive mesh refinement in space-time.
Mathematical Formulation
Definition of operators L and M.
Variational problem and function spaces defined.
Discrete Variational Problem
Derivation of discrete variational problem using GLS stabilization.
Analysis of Stabilized Formulations
Boundedness and coercivity analysis for stability.
Proof of unique solution existence based on bilinear form properties.
Error Analysis
A priori error estimates derived for GLS stabilized advection-diffusion equation.
A posteriori error indicator calculation for adaptive refinement strategy.
Numerical Examples
Convergence study for heat equation with Gaussian pulse initial condition.
Comparison between sequential time-marching and space-time solutions for advection-diffusion equation with smooth initial condition.
Adaptive Solutions
Adaptive refinement behavior demonstrated for heat diffusion problem.
Adaptive refinement results shown for advection-diffusion problem with smooth initial condition.
Conclusion
Summary of findings regarding convergence, stability, efficiency, and accuracy achieved through space-time finite element analysis.
Stats
The diffusivity value ν is fixed at 10^-4 (Section: Introduction).
The global Peclet number is given as Peg ≈4.4×10^8 (Section: Adaptive Solutions).
Quotes
"In this work, we tackle two key aspects associated with solving evolution equations in space-time – stability and computational cost."
"The idea of parallelism in both space and time builds on a rich history of parallel time integration."