Numerical
Analysis of Partial
Differential Equations
MA
8445-6
Fall 2007/Spring 2008
- Objective: This course is intended to be an introduction to the mathematical analysis
of finite element (and some finite volume) methods for partial
differential equations. The distinctive features of this course are
that
- it uses
the basic concepts of continuous dependence, a priori and posteriori
error estimates and duality arguments,
- deals with the issues of
computational complexity and adaptivity, and
- emphasizes the interplay between theory
and practice.
- References:
- D. Braess, Finite
Elements. Theory, Fast Solvers and Applications in Solid Mechanics
- F. Brezzi and M. Fortin, Mixed and hybrid finite element methods
- P.G. Ciarlet, The
Finite Element Method for Elliptic Problems
- Y. Saad, Iterative
Methods for Sparse linear systems
- C. Johnson, Numerical
solutions of PDEs by the Finite Element Method
- R. LeVeque, Numerical
methods for conservation laws
- R. LeVeque, Finite
Volume Methods for Hyperbolic Problems
- A. Quarteroni and A. Valli, Numerical methods for PDEs
- Syllabus:
- For the Fall:
- The two-point
boundary value problem
- Examples and their physical interpretation
- The weak formulation and the equivalent energy minimization
problem
- The continuous Galerkin method: Existence and uniqueness of
the approximation
- The matrix equations
- Lagrange multipliers and static condensation
- A priori error estimates in the energy norm
- Approximation theory
- Algebraic, spectral and exponential convergence
- A priori error estimates in the L2-norm: The Aubin-Nitsche
duality argument
- Superconvergence at the nodes for both the potential and its
flux
- A priori estimates of the error in negative-order norms
- Approximation of linear functionals
- Postprocessing by local convolution
- A posteriori estimates in the energy norm
- An adaptive method and its convergence
- Second-order
elliptic problems in several space dimensions
- Examples and their physical interpretation
- The general abstract approach
- The Lax-Milgam Lemma
- The Poincare and trace inequalities: Density and continuity
- Finite-dimensional subspaces of H1: Distributional
derivatives
- The continuous Galerkin method
- Definition of a finite element method
- Examples of finite element methods
- The matrix equations: Implementation details
- Lagrange multipliers and static condensation
- A priori error estimates in the energy norm
- A priori error estimates in the L2-norm: The Aubin-Nitsche
duality argument
- Elliptic regularity results
- A priori error estimates in negative-order norms
- Postprocessing by local convolution
- A posteriori error estimates
- An adaptive method and its convergence
- Methods
for solving the matrix equation
- Direct methods
- Classic iterative methods
- The method of steepest descent
- Error estimates: The Kantorovich inequality
- The method of conjugate gradients
- Error estimates: Chebyshev polynomials
- A mixed method
- The Raviart-Thomas (RT) method
- Existence and uniqueness of the approximation
- The inf-sup condition: Characterization of surjective
operators
- The commuting-diagram property
- A priori error estimates
- Hybridization of the method
- Superconvergence of the Lagrance multipliers
- Local postprocessing of the approximate solution
- For the Spring:
- Convection-diffusion
problems
- The model convection-diffusion problem
- Continuous Galerkin methos: The need of stabilization
- Streamline-diffusion methods
- Discontinuous Galerkin methods
- The
GMRES method
- The conjugate gradient method as a Galerkin method
- Arnoldi's orthogonalization
- The full orthogonalization method
- The GMRES method
- Convergence analysis
- Convection-diffusion
problems
- The model convection-diffusion problem
- Continuous Galerkin methos: The need of stabilization
- Streamline-diffusion methods
- Discontinuous Galerkin methods
- Linear
hyperbolic conservation laws
- The discontinuous Galerkin space discretization
- A priorir error analysis of the semidiscrete method
- The RKDG method
- Stability and accuracy of the method
- Amplitude and phase errors
- Friedrich's systems
- Non-linear
scalar hyperbolic conservation laws
- The loss of well-posedness
- Traveling wave solutions: The entropy condition
- The Riemann problem
- The entropy solution
- Continuous dependence results
- Monotone schemes
- Convergence and error estimates
- The RKDG method
- Compressible
fluid flow
- The Euler equations for gas dynamics
- The RKDG method
- The compressible Navier-Stokes equations
- Incompressible
fluid flow
- The Stokes, Oseen and incompressible Navier-Stokes equations
- Mixed methods for the Stokes equations
- Discontinuous Galerkin methods for the Stokes equations
- Obtaining exactly incompressible velocity approximations
- Discontinuous Galerkin methods for the Oseen equations
- Discontinuous Galerkin methods for the Navier-Stokes equations