About this Page 

This page is maintained by Peter Hellekalek.
On this page, you will find information on RNGs in stochastic simulation
but, not yet, on cryptographical applications.
You might want to study last year's news on the
old page. 



Automatic Nonuniform
Random Variates
(February 2004)


Wolfgang Hörmann, Josef Leydold, and
Gerhard Derflinger (Univ. of Economics, Vienna) have published a new
book on nonuniform random variate generation, see this
link to amazon.com. 
Empirical
Evidence concerning AES
(December 2002) 

pLab members Stefan
Wegenkittl and
Peter Hellekalek have tested AES, the advanced encryption standard,
in random generator modes. The results were convincing, the paper
is available
online from our ftp server. This paper also contains a discussion
of Maurer's Universal Statistical Test. 
Entropy
(June 2002) 

pLab member Stefan
Wegenkittl has uploaded the
slides of his talk at the Workshop
on Random Number Generators and Highly Uniform Point Sets at the
University of Montreal, 1728 June, 2002. Stefan's talk focuses upon
Maurer's Universal Randomness Test. 
Workshop
(May 2002) 

Pierre L'Ecuyer is organizing
a Workshop on Random
Number Generators and Highly Uniform Point Sets at the University
of Montreal, 1728 June, 2002. This workshop will unite many of the
leading specialists in those two fields. (See our LINKS
page for a list of webpages and/or email addresses of these collegues)

Improved ran_array
Initialization
(February 2002) 

Knuth's famous generator ran_array has
been updated. You can read part of the story at the
Recent News page of Donald Knuth as well as get the new
code from there. ran_array showed similar deficiencies to those
that the new Mersenne Twister initialization tries to cure. Pedro
Gimeno and Richard Brent have also been involved in this effort.

Improved
Mersenne Twister Initialization
(February 2002) 

Makoto
Matsumoto has improved the initialization routine of his famous
Mersenne Twister random number generator, follow this
link for further information. This 'update' corrects a minor problem
that has been described first by Jeff
Szuhay of
Psychology Software Tools in the context of simulations using
TT800, see the cautionary tale about
reseeding random number generators. 
Important Conference
(January 2002) 

The series of the MCQMC conferences continues
with MCQMC2002
in Singapore, Nov. 2528, 2002. This international conference will
be organized by
Harald Niederreiter (chair) and his collegues at National University
of Singapore. The MCQMC conferences
are a "must" for anyone interested in random number generation and/or
quasiMonte Carlo methods. 
Literature
on RNGs
(January 2002) 

I have updated the LITERATURE
page on this server and added a section on cryptographical RNGs, see
Literature on Random Number
Generators. 
Spectral Test
(December 2001) 

Within my research projects NB7576 and
FWFS8303, we have documented and implemented the most important practical
test for linear random number generators, the spectral test. Our Spectral
Test Server is available in a betaversion. This server is being
constructed by Karl Entacher and Peter Hellekalek 
Nonuniform
Random Numbers
(September 2001) 

In addition to their most
helpful collection of fast and reliable collection of algorithms for
nonuniform random numbers
UNURAN, the ARVAG Team has now presented
ANURAN, a project to develop and implement automatic code
generators for nonuniform random number generation. The goal
of this project is that programmers (that might have little or no
experience with random number generation) can produce source code
for fast, reliable, high quality random variate generators for large
classes of distributions which are not necessarily standard distributions.

Spectral Test Server
Down
(September 2001) 

pLab is sorry to announce that our new
Spectral
Test Server is down for much longer than expected because of a
hardware problem. This server had been constructed by Karl Entacher and Peter Hellekalek. It will take us some time
to get this thing running again. 
Safe
Sample Sizes
(September 2001) 

Makoto
Matsumoto (see also his page at
http://www.math.keio.ac.jp/~matumoto/eindex.html) has developed
a clever technique to determine safe and risky sample sizes for random
number generators. These results are of interest to practitioners:
we now know for several RNGs (like the UNIX "random" function, or
RANLUX of Lüscher, etc.) which sample sizes are likely to give
correct results in simulations. His papers will appear in the MCQMC2000
proceedings (Springer Verlag), the results on RANLUX have been presented
at the MCM2001. 
Entropy Measures
(September 2001) 

Stefan
Wegenkittl has compared
Ueli Maurer's "Universal Statistical Test for Random Bit Generators"
to the wellknown family of serial tests. Stefan was able to find
interesting relations and provides further insight into empirical
(statistical) testing of RNGs. 
RNGs
for QuasiMonte Carlo
(September 2001) 

Karl
Entacher has used widely available LCGs to generate lowdiscrepancy
point sets for quasiMonte Carlo integration, see his new
page. This new approach to employ RNGs not only for Monte Carlo
but also for quasiMonte Carlo purposes was also taken up by Pierre
L'Ecuyer and Christiane
Lemieux, see their recent joint publications.

AES Implementation
in C++
(May 2001) 

pLab member Gerhard
Wesp has implemented Rijndael, in other words, the Advanced Encryption
Standard AES,
in C++. The implementation is available via the Rijndael
home page 
Exploring
Randomness
(April 2001) 

Gregory J. Chaitin is
exploring randomness again, in his recent book published by SpringerVerlag,
London. I would recommend anyone interested in randomness to visit
Gregory's
homepage and study his numerous publications on the limits . See
also the
online article in New
Scientist. 
Nonuniform Random
Numbers
(April 2001) 

Josef Leydold, Wolfgang Hoermann, Erich
Janka, and Günter Tirler of the University of Economics Vienna
have written a library of functions to generate nonuniform
random numbers called UNURAN
which reflects the stateoftheart in this field. UNU.RAN is an ANSI
C library licensed under GPL. It contains universal (also called automatic
or blackbox) algorithms that can generate random numbers from large
classes of continuous or discrete distributions, and also from practically
all standard distributions. 
Parameters
for ICGs
(March 2001) 

In order to implement
inversive congruential generators (ICGs), one needs to know socalled
IMPpolynomials. Their coefficients will be the parameters for the
ICGs. Tables of such polynomials and, hence, tables of parameters
for ICGs are availabe from the Winter
Simulation Conference 1995 survey paper by Peter Hellekalek and
in the form of extensive tables
of IMPpolynomials. 
Conference
(February 2001) 

The 3rd IMACS Seminar on Monte Carlo Methods
MCM2001 will take place at Salzburg
University from September 1014, 2001. A session on random number
generation will be organized by Peter Hellekalek. As invited speakers, Luc Devroye
(Montreal), Karl Entacher (Salzburg), Pierre L'Ecuyer (Montreal),
and Ueli Maurer (ETH Zurich) have already agreed to give a talk. We
will cover the whole range from uniform RNGs for Monte Carlo simulation
(L'Ecuyer), via nonuniform random numbers (Devroye) to random numbers
in cryptography (Maurer) and the relation to quasiMonte Carlo methods
(Entacher). 
Spectral
Test Server
(August 2000) 

pLab is proud to announce
our new Spectral Test
Server. This server allows you to find good LCGs (linear congruential
generators) by the spectral test, an important figure of merit for
random number generators. This server has been constructed by
Karl Entacher and
Peter Hellekalek. 
A Cautionary Tale
from Practice
(August 2000) 

Jeff
Szuhay of
Psychology Software Tools has provided a cautionary
tale about reseeding random number generators. He experienced
problems when reseeding the excellent generator TT800 of Makoto
Matsumoto in field use. Anyone interested in further details is
asked to contact Jeff directly. 
MCQMC
2000, Hongkong
(August 2000) 

The
Fourth International Conference on Monte Carlo and QuasiMonte Carlo
Methods in Scientific Computing in Hongkong, Nov. 27  Dec. 3,
2000, will unite most of the leading specialists in these fields.
It is the most important meeting in this field and takes place every
two years. This series of conferences is closely connected with Harald
Niederreiter, who has initiated this series of conferences with
his 1990 Alaska lectures. All conference proceedings have been published
by SpringerVerlag. 