PyMT64 1.5

Python version of the Mersenne Twister 64-bit pseudorandom number generator

PyMT64 Package

                 PyMT64

PyMT64 is a Python version of the Mersenne Twister (MT) 64-bit pseudorandom number generator by Takuji Nishimura and Makoto Matsumoto (see http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt64.html and the references below).

This customised version is thread safe and was interfaced from C to Python (see pymt64.c)

This module provides the following methods:

• init : initialization of the state vector (mt) used by the pseudorandom number generator (PNG)
• uniform : generation of an uniform distribution
• normal : generation of two Normal distributions
• poisson : generation of a Poisson distribution

The period of the PNG is 2**19937-1.

Example: import time import pymt64 seed = int(time.time()) # the initial seed mt = pymt64.init(seed) # initialisation of the state vector of MT u = pymt64.uniform(mt,10) # generation of an uniform distribution print u [ 0.94295421 0.47222327 0.634552 0.26012686 0.38431784 0.23995444 0.02175826 0.8209848 0.79266556 0.8638286 ]

For a complete example, see pymt64_test.py

Note: the state vector 'mt' returned by pymt64.init has 313 elements instead of the 312 elements of the original C code. This is because the 313th element store the associated counter (mti).

Change history: 1.5 : module interface is now based on Cython, module now compatibily with python 3 1.4 : link problem fixed 1.3 : fixe a compilation problem regading the Numpy include directory 1.2 : the previous implementation of the poisson distribution was not thread safe 1.1 : fix a problem with the initialization of the seed (in the previous version the seed set by init() was not taken into account such that the results were not reproductible) 1.0 : initial version

R. Samadi (LESIA, Observatoire de Paris), 22 Dec. 2012

References: T. Nishimura, Tables of 64-bit Mersenne Twisters'' ACM Transactions on Modeling and Computer Simulation 10. (2000) 348--357. M. Matsumoto and T. Nishimura, Mersenne Twister: a 623-dimensionally equidistributed uniform pseudorandom number generator'' ACM Transactions on Modeling and Computer Simulation 8. (Jan. 1998) 3--30.