Mathematics

Advanced Numerical Analysis

Module code: 852G1
Level 7 (Masters)
15 credits in autumn teaching
Teaching method: Lecture
Assessment modes: Coursework, Unseen examination

This module will cover topics including:

  • Iterative methods for linear systems: Jacobi and Gauss-Seidel, conjugate gradient, GMRES and Krylov methods
  • Iterative methods for nonlinear systems: fixed point iteration, Newton's method and Inexact Newton
  • Optimisation: simplex methods, descent methods, convex optimisation and non-convenx optimisation
  • Eigenvalue problems: power method, Von Mises method, Jacobi iteration and special matrices
  • Numerical methods for ordinary differential equations: existence of solutions for ODE's, Euler's method, Lindelöf-Picard method, continuous dependence and stability of ODE's
  • Basic methods: forward and backward Euler, stability, convergence, midpoint and trapezoidal methods (order of convergence, truncation error, stability convergence, absolute stability and A-stability)
  • Runge-Kutta methods: one step methods, predictor-corrector methods, explicit RK2 and RK4 as basic examples, and general theory of RK methods such as truncation, consitency, stability and convergence 
  • Linear multistep methods: multistep methods, truncation, consistency, stability, convergence, difference equaitons, Dahlquist's barriers, Adams family and backward difference formulas
  • Boundary value problems in 1d, shooting methods, finite difference methods, convergence analysis, Galerkin methods and convergence analysis

Module learning outcomes

  • Analyse in depth the convergence properties of advanced iterative methods
  • Implement and apply advanced iterative methods to solve linear and nonlinear problems
  • Analyse in depth the convergence and stability properties of advanced time-stepping methods
  • Conduct thorough error analysis
  • Implement and apply time-stepping methods to solve ODE's and time-dependent PDE's including boundary value problems