STOCHASTIC OPTIMISATION METHODS

• Monte-Carlo method with interval reduction

• Genetic Algorithms

  • Mimic natural evolution process

• Simulated Annealing

  • Thermodynamical analogy of annealing of metals

• Merits

  • Non-linear, multi-modal, discontinuous functions
  • Non-Greedy optimisation------global minimum (?)
  • Better domain exploration

• Demerits

  • Tuning of optimisation parameters
  • Large number of function evaluations
  • Convergence

Previous slide

Next slide

Back to first slide

View graphic version