4th International ICST Workshop on Tools for solving Structured Markov Chains

Research Article

A Software Tool for the Steady–State Analysis of Google–like Stochastic Matrices

  • @INPROCEEDINGS{10.4108/ICST.VALUETOOLS2009.7453,
        author={Tugrul Dayar and Gokce Nil  Noyan},
        title={A Software Tool for the Steady--State Analysis of Google--like Stochastic Matrices},
        proceedings={4th International ICST Workshop on Tools for solving Structured Markov Chains},
        publisher={ACM},
        proceedings_a={SMCTOOLS},
        year={2010},
        month={5},
        keywords={Markov chains steady\{state vector Google},
        doi={10.4108/ICST.VALUETOOLS2009.7453}
    }
    
  • Tugrul Dayar
    Gokce Nil Noyan
    Year: 2010
    A Software Tool for the Steady–State Analysis of Google–like Stochastic Matrices
    SMCTOOLS
    ICST
    DOI: 10.4108/ICST.VALUETOOLS2009.7453
Tugrul Dayar1,*, Gokce Nil Noyan2,*
  • 1: Department of Computer Engineering, Bilkent University, TR06800 Bilkent, Ankara, Turkey.
  • 2: Undersecretariat for Defence Industries, Ziyabey Caddesi, 21 Sokak, No.4 TR06520 Balgat, Ankara, Turkey.
*Contact email: tugrul@cs.bilkent.edu.tr, gnnoyan@ssm.gov.tr

Abstract

The problem of computing the steady-state vector of positive stochastic matrices which are convex combinations of sparse, nonnegative matrices possibly having zero rows with appropriately chosen rank-1 matrices is addressed by a software tool. The dynamically changing matrices used by the Google search engine in ranking web pages are among the largest of such matrices. Ranking pages amounts to solving for the steady-state vectors of these matrices. The tool implements the power, quadratically extrapolated power, and iterative methods based on various block partitionings, including those with block triangular form and with triangular blocks obtained using cutsets.