Department of Mathematics
University of Pisa

 Numerical Analysis  and Computational Mathematics


Numerical Methods for Markov Chains

 


The research in this field concerns the design and analysis of numerical methods for solving Markov chains. Particular attention is devoted to structures arising from the modeling  of problems in queueing theory. The main problems investigated are related to the solution of M/G/1-type and G/M1-type Markov chains, QBD problems, PH/PH/1 queues. Efficiency of the algorithms in terms of their complexity and stability is pursued.

Collaborations:
 S. Chakravarthy, P. Favati, U. Krieger, G. Latouche, V. Ramaswami, N. Rhee

Papers:

Software:
 
  • Cyclic Reduction for QBD problems.
  • Point-wise Cyclic Reduction for M/G/1 Matrices.
  • Doubling Method for M/G/1 Matrices
  • Non-skip-free M/G/1 Matrices
  • PH/PH/1/N Queues

  •  

     

    Conferences:

    Third International Conference on Matrix Analytic Methods, Leuven, July 2000.

    Related Links
    Matrix Analytic Home Page