The use of non-linear programming in matrix analytic methods

In this paper, we present new methods for finding the R and G matrices which play a crucial role in determining the steady-state distribution of Markov chains of the GI/M/1 and M/G/l type respectively. The methods involve finding solutions to some non-linear programming problems. We obtain the Karush-Kuhn-Tucker (KKT) conditions for two of the non-linear programming problems and, for the M/G/l case, show that there is a simple relationship between the dual variables and the G matrix. We present two algorithms for solving the non-linear programming problems. We have also carried out numerical investigations for the M/G/l problem and found that one of the algorithms often performs much better than what we call the “standard method” of solution

[1]  Mokhtar S. Bazaraa,et al.  Nonlinear Programming: Theory and Algorithms , 1993 .

[2]  B. Sengupta.  Markov processes whose steady state distribution is matrix-exponential with an application to the GI/PH/1 queue , 1989, Advances in Applied Probability.

[3]  I. Olkin,et al.  Inequalities: Theory of Majorization and Its Applications , 1980 .

[4]  Marcel F. Neuts,et al.  Matrix-Geometric Solutions in Stochastic Models , 1981 .

[5]  Winfried K. Grassmann,et al.  Equilibrium distribution of block-structured Markov chains with repeating rows , 1990, Journal of Applied Probability.

[6]  J. Ortega.  Numerical Analysis: A Second Course , 1974 .

[7]  G. Latouche.  ALGORITHMS FOR INFINITE MARKOV CHAINS WITH REPEATING COLUMNS , 1993 .

[8]  Marcel F. Neuts,et al.  Matrix-geometric solutions in stochastic models - an algorithmic approach , 1982 .

[9]  Levent Gun,et al.  Experimental results on matrix-analytical solution techniques–extensions and comparisons , 1989 .

[10]  Vaidyanathan Ramaswami,et al.  An experimental evaluation of the matrix-geometric method for the GI/PH/1 queue , 1989 .

[11]  K. Sohraby,et al.  An invariant subspace approach in m/g/l and g/m/l type markov chains , 1997 .

[12]  Latouche Guy,et al.  A note on two matrices occurring in the solution of quasi-birth-and-death processes , 1987 .

[13]  E. M. L. Beale,et al.  Nonlinear Programming: A Unified Approach. , 1970 .

[14]  Vaidyanathan Ramaswami,et al.  A logarithmic reduction algorithm for quasi-birth-death processes , 1993, Journal of Applied Probability.

[15]  Vaidyanathan Ramaswami,et al.  Nonlinear Matrix Equations in Applied Probability—Solution Techniques and Open Problems , 1988 .

[16]  Edward P. C. Kao.  Using State Reduction for Computing Steady State Probabilities of Queues of GI/PH/1 Types , 1991, INFORMS J. Comput..