Skip to main content
Skip header
Terminated in academic year 2009/2010

Methods of Optimization

Type of study Follow-up Master
Language of instruction Czech
Code 457-0313/02
Abbreviation MO
Course title Methods of Optimization
Credits 6
Coordinating department Department of Applied Mathematics
Course coordinator prof. RNDr. Zdeněk Dostál, DSc.

Subject syllabus

Lectures:
An introduction to the calculus of variations. Linear spaces, funkcionls and their differentials (Fréchet, Gateaux).
Euler equation and the solution of the classical problems of variational calculus.
Unconstrained minimization. One-dimensional minimization of unimodular functions.
Conditions of minimum, the Newton method and its modification. Gradient methods, method of conjugate gradients.
Constrained minimization. Karush-Kuhn-Tucker conditions of optimality.
Penalization and barrier methods for constrained minimization. Feasible direction method (SLP) and active set strategy for bound constrained problems.

Duality in convex programming. Saddle points, Uzawa algorithm and augmented Lagrangians.
Linear programming, simplex method.
Non-smooth optimization, subgradients and optimality conditions.
Global optimization, genetic and evolutionary algorithms, simulated annealing, tabu search.
Software.

Exercises:
Introduction to the MATLAB programming.
Implementation of the golden section and Fibonacci series methods.
Implemenation of the Newton-like methods.
Implementation of the gradient based method.
Implementation of the conjugate gradient method.

Implementation of the penalty methody for equality constrained minimization.
Implementation of the feasible direction method (SLP).
Implementation of the active set method for bound constrained quadratic programming.
Implementation of the augmented Lagrangian metod.
Implementation of algorithms for global optimization.
Solution of selected engeneering problems using optimization software.

Projects:
Comparing performance of the methods for unconstrained optimization using a numerical example (max 10 marks).
Comparing performance of the methods for constrained optimization using a numerical example (max 10 marks).
Solution of a selected engineering problem (max 10 marks).


Computer labs:

Introduction to the MATLAB programming.
Implementation of the golden section and Fibonacci series methods.
Implemenation of the Newton-like methods.
Implementation of the gradient based method.
Implementation of the conjugate gradient method.
Implementation of the penalty methody for equality constrained minimization.
Implementation of the feasible direction method (SLP).
Implementation of the active set method for bound constrained quadratic programming.
Implementation of the augmented Lagrangian metod.

Implementation of algorithms for global optimization.
Solution of selected engeneering problems using optimization software.

Literature

D. P. Bertsekas, Nonlinear Programming, Athena Scientific, Belmont 1999. ISBN 1-886529-00-0.
M. S: Bazaraa, C. M. Shetty, Nonlinear programming, J. Wiley, New York 1979, ruský překlad Mir Moskva 1982.
R. Fletcher, Practical Methods of Optimization, John Wiley & sons,Chichester 1997.

Advised literature

D. T. Pham and D. Karaboga, Intelligent Optimization Techniques, Springer, London 2000. ISBN 1-85233-028-7.

Z. Dostal, Optimal Quadratic Programming Algorithms: With Applications to Variational Inequalities Springer, New York 2009. ISBN: 0387848053 , ISBN-13: 9780387848051