• DocumentCode
    185273
  • Title

    Uniformity and correlation test parameters for random numbers generators

  • Author

    Petrila, Iulian ; Manta, Vasile ; Ungureanu, Florina

  • Author_Institution
    Fac. of Autom. Control & Comput. Eng., “Gheorghe Asachi” Tech. Univ. of Iasi, Iasi, Romania
  • fYear
    2014
  • fDate
    17-19 Oct. 2014
  • Firstpage
    231
  • Lastpage
    236
  • Abstract
    In this study we propose a series of test parameters for random number generators in order to analyze the uniformity and the degree of correlation of generated numbers. The present analysis was applied to a series of standard generators used in libraries of the usual programming languages such as: C/C++, Java and Fortran. Besides using some standard tests, new parameters that provide a more adequate description of uniformity and correlation are proposed. The analysis attempts to explain the different results obtained in simulations performed with different types of generators and to outline the different degree of uniformity and correlation of the numbers produced by these generators. The uniformity and correlation properties were analyzed both independently and together through integrants parameters. The analysis revealed that high sizes of storage and representation or high periodicity of the random number generators does not offer at the same time both high uniformity and less correlation of random numbers packets with different sizes.
  • Keywords
    random number generation; software libraries; C language; C++ language; Fortran language; Java language; correlation test parameters; programming languages; random number generators; software libraries; Correlation; Correlation coefficient; Generators; Java; Libraries; Standards; Vectors; linear methods; pseudorandom numbers; random number generation; uniformity and correlation tests;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    System Theory, Control and Computing (ICSTCC), 2014 18th International Conference
  • Conference_Location
    Sinaia
  • Type

    conf

  • DOI
    10.1109/ICSTCC.2014.6982421
  • Filename
    6982421