Telek, Miklós (BME HIT)
Concentrated
matrix exponential distributions and their application in transient analysis of
stochastic processes and inverse Laplace transformation
Concentrated distributions are important for modeling deterministic delays in stochastic models. In Markovian environments the Erlang distribution of order $n$, whose squared coefficient of variation is $1/n$, is the most concentrated distribution which can be obtained by a Markov chain of $n+1$ states.
Relaxing the sign constraints of Markovian generator matrices more concentrated distributions (referred to as matrix exponential distributions) can be obtained. We present a class of distributions with $O(1/n^2)$ squared coefficient of variation.
Potential application of such Concentrated matrix exponential distributions in transient analysis of Markovian models and Inverse Laplace transformation is also presented.
The talk is
held in English!
Az előadás
nyelve angol!
Date: Oct 27, Tuesday 4:15pm
Place: BME, Building „Q”, Room
QBF13