Article

A comparative study of some pseudorandom number generators

Research Institute for Theoretical Physics, P.O. Box 9 (Siltavuorenpenger 20 C), University of Helsinki, FIN-00014 Helsinki, Finland
Computer Physics Communications (Impact Factor: 3.11). 04/1993; 86(3):209-226. DOI: 10.1016/0010-4655(95)00015-8
Source: arXiv

ABSTRACT

We present results of a test program of a group of pseudorandom number generators which are commonly used in the applications of physics, in particular in Monte Carlo simulations. The generators include public domain programs, manufacturer installed routines and a random number sequence produced from physical noise. We start by traditional standard tests, followed by detailed bit level and visual tests. The computational speed of various algorithms is also scrutinized. Our results allow direct comparisons between the properties of different generators, as well as an assessment of the efficiency of various test methods. Together with recently developed application specific tests, this information provides a good criterion to choose the best generator among the tested ones for a given problem.

Download full-text

Full-text

Available from: Ilpo Vattulainen, Aug 14, 2014
  • Source
    • "multi-random numbers). Unfortunately, most of the existing random number generators are virtually unable to meet these requirements: the correlation between the generated sequences by a classical PNRG increases with the number of sequences [12]. The authors in [10] analyze popular PRNGs working in parallel (e.g. the 32-bit generator rand, the combined multiple-recursive generator, and the additive lagged-Fibonacci generator), and demonstrate the lack of independence of the generated streams. "
    [Show abstract] [Hide abstract]
    ABSTRACT: An improved Lozi function with alternate coef-ficients has been proposed. The modifications in the model allow to remove the holes in the attractor which are not desir-able, but appeared in the previous Lozi function; in this way, an everywhere dense attractor can be obtained. Moreover, the strong sensitivity to the type of binarisation (conversion of real values to 0 and 1) has been demonstrated; this conversion to binary numbers is instrumental to apply the NIST tests for randomness. The results have been validated and compared via NIST tests, for the different methods of quantization. Fi-nally, is has been verified that the multi-random properties of the output signal have been improved thanks to the following strategies : under-sampling of the output signal, and the sys-tem order increasing.
    Full-text · Dataset · Jan 2013
  • Source
    • "multi-random numbers). Unfortunately, most of the existing random number generators are virtually unable to meet these requirements: the correlation between the generated sequences by a classical PNRG increases with the number of sequences [12]. The authors in [10] analyze popular PRNGs working in parallel (e.g. the 32-bit generator rand, the combined multiple-recursive generator, and the additive lagged-Fibonacci generator), and demonstrate the lack of independence of the generated streams. "
    [Show abstract] [Hide abstract]
    ABSTRACT: An improved Lozi function with alternate coefficients has been proposed. The modifications in the model allow to remove the holes in the attractor which are not desirable, but appeared in the previous Lozi function; in this way, an everywhere dense attractor can be obtained. Moreover, the strong sensitivity to the type of binarisation (conversion of real values to 0 and 1) has been demonstrated; this conversion to binary numbers is instrumental to apply the NIST tests for randomness. The results have been validated and compared via NIST tests, for the dfferent methods of quantization. Finally, is has been verified that the multi-random properties of the output signal have been improved thanks to the following strategies : under-sampling of the output signal, and the system order increasing.
    Full-text · Article · Jan 2013
  • Source
    • "Rule1 and rule2 are alternatively used in this process, and they can be designed in such a way that a certain degree of randomness can be incorporated. For example, rule1 can be designed using linear feedback shift register (LFSR) with maximum length [13]. Rule2 simply bitwise inverts the address generated by rule1. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Flying-adder architecture is an emerging on-chip frequency synthesis technology. Its theoretical foundation is the time-average-frequency concept introduced in 2008. During the past decade, the flying-adder technique has been successfully used in numerous commercial products. However, when fractional control word is used, its clock spectrum contains spurious tones due to the periodic carries generated from the fractional accumulation. In this brief, a storage-based technique is introduced to randomize the carries so that the discrete spurs can be converted into noise.
    Full-text · Article · Jun 2011 · Circuits and Systems II: Express Briefs, IEEE Transactions on
Show more