shuuji3.xyz / python-extension-twister
[READ-ONLY] Mirror of https://github.com/shuuji3/python-extension-twister. Experimental Python C extension to generate random numbers by using Mersenne twister
0
Fork 0

Configure Feed

Select the types of activity you want to include in your feed.

python-extension-twister / src / mt19937ar / readme-mt.txt
3.0 kB 79 lines
1This is a Mersenne Twister pseudorandom number generator 2with period 2^19937-1 with improved initialization scheme, 3modified on 2002/1/26 by Takuji Nishimura and Makoto Matsumoto. 4modified on 2005/4/26 by Mutsuo Saito 5 6Contents of this tar ball: 7readme-mt.txt this file 8mt19937ar.c the C source (ar: initialize by ARray) 9mt19937ar.h the C header file for mt19937ar 10mtTest.c the C test main program of mt19937ar.c 11mt19937ar.out Test outputs of six types generators. 1000 for each 12 131. Initialization 14 The initialization scheme for the previous versions of MT 15(e.g. 1999/10/28 version or earlier) has a tiny problem, that 16the most significant bits of the seed is not well reflected 17to the state vector of MT. 18 19This version (2002/1/26) has two initialization schemes: 20init_genrand(seed) and init_by_array(init_key, key_length). 21 22init_genrand(seed) initializes the state vector by using 23one unsigned 32-bit integer "seed", which may be zero. 24 25init_by_array(init_key, key_length) initializes the state vector 26by using an array init_key[] of unsigned 32-bit integers 27of length key_kength. If key_length is smaller than 624, 28then each array of 32-bit integers gives distinct initial 29state vector. This is useful if you want a larger seed space 30than 32-bit word. 31 322. Generation 33After initialization, the following type of pseudorandom numbers 34are available. 35 36genrand_int32() generates unsigned 32-bit integers. 37genrand_int31() generates unsigned 31-bit integers. 38genrand_real1() generates uniform real in [0,1] (32-bit resolution). 39genrand_real2() generates uniform real in [0,1) (32-bit resolution). 40genrand_real3() generates uniform real in (0,1) (32-bit resolution). 41genrand_res53() generates uniform real in [0,1) with 53-bit resolution. 42 43Note: the last five functions call the first one. 44if you need more speed for these five functions, you may 45suppress the function call by copying genrand_int32() and 46replacing the last return(), following to these five functions. 47 483. main() 49main() is an example to initialize with an array of length 4, 50then 1000 outputs of unsigned 32-bit integers, 51then 1000 outputs of real [0,1) numbers. 52 534. The outputs 54The output of the mt19937ar.c is in the file mt19937ar.out. 55If you revise or translate the code, check the output 56by using this file. 57 585. Cryptography 59This generator is not cryptoraphically secure. 60You need to use a one-way (or hash) function to obtain 61a secure random sequence. 62 636. Correspondence 64See: 65URL http://www.math.keio.ac.jp/matumoto/emt.html 66email matumoto@math.keio.ac.jp, nisimura@sci.kj.yamagata-u.ac.jp 67 687. Reference 69M. Matsumoto and T. Nishimura, 70"Mersenne Twister: A 623-Dimensionally Equidistributed Uniform 71Pseudo-Random Number Generator", 72ACM Transactions on Modeling and Computer Simulation, 73Vol. 8, No. 1, January 1998, pp 3--30. 74 75------- 76Copyright (C) 1997 - 2002, Makoto Matsumoto and Takuji Nishimura, 77All rights reserved. 78Copyright (C) 2005, Mutsuo Saito 79All rights reserved.