Overall Course Objectives
Numerical simulation is used for solving problems which are so complex that a theoretical model of the system cannot be solved by analytical methods. The goal is to enable the student to formulate a model of a real-life problem, implement and validate this model on a computer, and perform experiments with the computer model. Emphasis is put upon models involving stochastic elements, which are simulated by so-called Monte Carlo methods.
- Apply build-in random number generators in various software products
- Implent algorithms for random number generation from various distributions
- Perform simple statistical analysis of simulated data
- Apply simulation based statistical techniques like bootstrap and Markov Chain Monte Carlo
- Apply simulated annealing as a heuristic method for minor optimization problems with discrete variables
- Develop and implement simulation procedures for simple technical systems using the event by event principle
- Perform verification of a programme for simulation
- Perform validation of a simulation model
- Plan and execute a simulation study for a specific part or function of some (technical) system
- Apply techniques for variance reduction in a simulation study
- Present the results of a simulation study writtenly or verbally
The first six to seven days deals with theory and exercises, followed by a case-study during the remaining part of the last two weeks. Attendance is mandatory for the first period of the course.
Theory: The modelling process, methods of solution, random number generation, sampling from statistical distributions. Discrete simulation: Time-true simulation, event-by-event principle, variance reduction. Case-studies: Operations research real-life problems, e.g. performance of data- and telecommunication systems, production planning, inventory control, optimal stochastic control etc.
02402, or a similar course in elementary statistics. Basic programming skills.
Lectures, exercises and project work.