The Fortran module RNDGEN
implements the KISS (Keep It Simple Stupid) random number generator as an object, facilitating the usage of multiple and independent generators.
The code adaptation is derived from Thomas Vojta's implementation available at http://web.mst.edu/~vojtat/class_5403/kiss05/rkiss05.f90.
For reference, the original source code can be found here, with the following information:
! Random number generator KISS05 after a suggestion by George Marsaglia
! in "Random numbers for C: The END?" posted on sci.crypt.random-numbers
! in 1999
!
! version as in "double precision RNGs" in sci.math.num-analysis
! http://sci.tech-archive.net/Archive/sci.math.num-analysis/2005-11/msg00352.html
!
! The KISS (Keep It Simple Stupid) random number generator. Combines:
! (1) The congruential generator x(n)=69069*x(n-1)+1327217885, period 2^32.
! (2) A 3-shift shift-register generator, period 2^32-1,
! (3) Two 16-bit multiply-with-carry generators, period 597273182964842497>2^59
! Overall period > 2^123
!
!
! A call to rkiss05() gives one random real in the interval [0,1),
! i.e., 0 <= rkiss05 < 1
!
! Before using rkiss05 call kissinit(seed) to initialize
! the generator by random integers produced by Park/Millers
! minimal standard LCG.
! Seed should be any positive integer.
!
! FORTRAN implementation by Thomas Vojta, [email protected]
! built on a module found at www.fortran.com
!
!
! History:
! v0.9 Dec 11, 2010 first implementation
! V0.91 Dec 11, 2010 inlined internal function for the SR component
! v0.92 Dec 13, 2010 extra shuffle of seed in kissinit
! v093 Aug 13, 2012 changed inter representation test to avoid data statements
Copy the file src/rndgen.f90
or add the package as a dependency using the Fortran Package Manager (Fpm):
[dependencies]
rndgen-fortran.git = "https://github.com/wcota/rndgen-fortran"
Import the module using use rndgen_mod
, and/or use rndgenPL_mod
for the power-law number generator.
Create the generator object by using
type(rndgen) :: generator
The generator requires a positive integer as a seed to initialize the generator with Park/Millers minimal standard LCG:
integer :: seed = 294727492
call generator%init(seed)
To generate a random number, use:
generator%rnd()
for a real number in the range [0,1)generator%int(i1, i2)
for an integer number in the range [i1, i2]generator%real(r1, r2)
for a real number in the range [r1, r2)
Reset the generator (to start again with the same sequence) with generator%reset()
.
It is possible to save and read the generated seeds. For that, an rndSeed
object needs to be declared as
type(rndSeed) :: seeds
To save, use call generator%save_seed(seeds, file_unit)
, where file_unit
refers to an opened file unit. It will save the current seed in the object and also write it to the file unit.
To read from a rndSeed
object, use call generator%read_seed(seeds)
, or to read from a file unit, use call generator%read_seed(seeds, file_unit)
.
The module rndgenPL_mod
extends the generator to an integer power-law distribution. The code was adapted from Silvio C. Ferreira's codes.
Declare the object with type(rndgenPL) :: generatorPL
, and initialize it with call generatorPL%initPL(kmin, kmax, gamma, seed)
, allowing the generation of random numbers following a power-law
To generate the number, use generatorPL%rndPL()
.
Using Fpm, execute:
fpm run --example example
to run the first example. The list of examples is as follows:
example
: Generates 10 random numbers between 0 and 1, integers between 5 and 2587, and floats between -5.2 and 100.9, resets, and repeats the process.example-vojta
: Original example by Thomas Vojta, from https://web.mst.edu/vojtat/class_5403/kiss05/rtest.f90example-save
: Generates 10 random numbers, resets, and saves the state after 5 runs. Then, reads from the file.example-2gen
: Runs two generators simultaneously. Usage:fpm run --example example-2gen -- seed1 seed2
example-2gen-invert
: Same as the previous example, but swaps the seeds after a reset.example-PL
: Generates four sequences of power-law distributed random numbers.
Expected outputs are available at example/output-*.txt
.
Tested with gfortran
, ifort
, and ifx
compilers. To use a specific compiler, run with fpm --compiler=ifort [...]
.