Thesis/Dissertation: Zuzana Komárková: Phase-type approximation techniques
Bachelor's thesis
Phase-type approximation techniques
Abstract
V této práci se zabýváme PH-aproximací, tedy aproximací libovolných pravděpodobnostních rozdělení pomocí rozdělení fázového typu. Popisujeme hlavní přístupy k tomuto problému a ke každému přístupu uvádíme řadu technik a již implementovaných nástrojů. Dále srovnáme tyto přístupy a stanovíme kritéria pro výběr vhodné metody v dané situaci.
Abstract
We consider the phase-type approximation problem, i.e. approximating general distributions by phase-type distributions. We describe the most common methods to deal with the problem. Further, for each method we survey the most relevant techniques and the corresponding tool support. Moreover, we compare the methods and discuss how to choose the most appropriate method under given circumstances.
Thesis description
Approximating general distributions by phase-type distributions is among popular techniques to analyze stochastic models, since the Markovian property of phase-type distributions allows analytical tractability. There is a large number of methods how to approximate given distributions by phase-type distributions. However, there is no work which summarizes such methods and their applicability in given stochastic model.
The task of the student is to study published approaches on the phase-type approximation and compare their suitability for specific models.
References:
- Osogami, T. and Harchol-Balter, M. A closed-form solution for mapping general distributions to minimal PH distributions. In Proc. 13th International Conference on Modelling Techniques and Tools for Computer Performance Evaluation, volume 2794 of LNCS, 200–217. Springer, 2003.
- A. Bobbio, A. Horváth, M. Scarpa, and M. Telek, Acyclic discrete phase type distributions: Properties and a parameter estimation algorithm, Performance Evaluation, vol. 54:1, pp. 1-32, 2003.
22/5/2012 09:40, doc. RNDr. Vojtěch Řehák, Ph.D., UČO 3721
Supervisor
Theses on a related topic
List of theses with an identical keyword.
-
Support vector machines for functional data
Mgr. Tomáš Pompa, UČO 500225 -
Model of demand for product of long-time usage
Bc. Lenka Potůčková -
FEM modelling of Dirac's equation
Mgr. Patrick Gono -
Processing of experimental data related to FGFR signalling pathways models
doc. Mgr. Ing Karel Sedlář, Ph.D. -
Collective risk models
Mgr. Veronika Veselá




