Year
2017
Units
4.5
Contact
3 x 50-minute lectures weekly
1 x 50-minute tutorial weekly
2 x 50-minute computer labs weekly
Enrolment not permitted
1 of MATH2722, MATH3701 has been successfully completed
Assumed knowledge
Working knowledge of first year mathematics and familiarity with basic linear algebra and differential equations.
Topic description
Representation of numbers, computer arithmetic, overflow and underflow, rounding error, cancellation error, truncation error, algorithms, stopping conditions. Solving f(x) = 0, bisection, secant, Newton methods. Interpolation, Lagrange polynomials, Hermite polynomials, divided differences, splines. Solving Ax = b, Gaussian elimination algorithm, pivoting strategies, LU decomposition, complexity, special matrices (diagonally dominant, positive definite, banded). Iterative methods, Jacobi, Gauss-Seidel, SOR methods. Eigenvalues and eigenvectors, power method, Householder's method, QR-algorithm. Non-linear systems, Newton's and steepest descent methods. Approximation theory, orthogonal polynomials, Legendre polynomials, Chebyshev polynomials. Numerical differentiation. Numerical integration, simple and composite rules, error estimates, Gaussian quadrature, adaptive quadrature methods, improper integrals, multiple integrals. Numerical methods in ODE, Runge-Kutta, stiff ODE's, predictor-corrector and adaptive techniques. Numerical solutions to partial differential equations.
Educational aims
  1. An understanding of the relationship between mathematical analysis of problems and the computation of numerical solutions
  2. An understanding of the sources of errors introduced by the use of computers in implementing mathematical descriptions of solutions
  3. An understanding of the role and methods of approximation
  4. An understanding of the methods for, and the limitations of, finding numerical solutions to problems
  5. Experience in scientific computing
  6. Experience in integrating mathematical derivations, numerical computations, figures and written text in presenting comprehensive solutions to problems
Expected learning outcomes
At the completion of this topic, students are expected to be able to:

  1. Understand the sources of errors introduced by the use of computers to perform computations
  2. Reformulate expressions so as to facilitate accurate computation
  3. Understand techniques for arriving at numerical solutions to many classes of problems including linear and non-linear systems of equations, integration, and differential equations
  4. Present solutions to problems by integrating mathematical derivations, numerical implementation, figures, and written text
  5. Understand both the breadth of problems to which the techniques can be applied and the limitations of solutions in individual circumstances
  6. Understand and implement numerical methods on their own