Speaker: Tommaso Nesti (University of Pisa)
Title: Numerical methods for computing the steady-state distribution of
a G-network.
Time: Wednesday, September 23; 16:00
Place: Sala seminari est, informatica
Abstract:
G-networks are a class of queueing networks introduced by E. Gelembe in
1989, which are characterized by the presence of positive and negative
customers. Unlike classical queueing networks, the equations yielding
the steady-state distribution of a G-network are nonlinear, resulting
in the need of developing efficient numerical methods for their solution.
In this talk we present two numerical methods for the computation of the
steady-state distribution, namely a fixed point and a Newton-Raphson
iteration. Using tools from nonnegative matrix theory, we prove that
both methods are locally convergent to the fixed point, with linear and
quadratic rate of convergence, respectively.
We compare these methods with an existing algorithm, concluding that the
Newton-Raphson iteration is a preferable choice for G-networks with a
moderate number of queues. We conclude by suggesting a possible
application for future research.
===
Everyone is welcome!
--
--federico poloni
Dipartimento di Informatica, Università di Pisa
http://www.di.unipi.it/~fpoloni/ tel. +39 050 2213143
