You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "Luc Maisonobe (JIRA)" <ji...@apache.org> on 2010/09/21 21:46:39 UTC
[jira] Created: (MATH-419) adding WELL random generators
adding WELL random generators
-----------------------------
Key: MATH-419
URL: https://issues.apache.org/jira/browse/MATH-419
Project: Commons Math
Issue Type: New Feature
Affects Versions: 2.1
Reporter: Luc Maisonobe
Assignee: Luc Maisonobe
Fix For: 2.2
Up to 2.1, we have few pseudo random number generators. We have an interface RandomGenerator implemented by three classes:
- JDKRandomGenerator that extends the JDK provided generator
- AbstractRandomGenerator as a helper for users generators
- BitStreamGenerator which in turn is extended only by MersenneTwister
The JDK provided generator is a simple one that can be used only for very simple needs. The Mersenne Twister is on the other hand a fast generator with good properties well suited for Monte-Carlo simulation. It is equidistributed for generating vectors up to dimension 623 and has a huge period: 219937 - 1.
Since Mersenne-Twister inception in 1997, some new generators have been created, retaining the good properties of Mersenne twister but removing some of its (few) drawbacks. The main one is that if initialized with a bits pool containing lots of zeroes, the pool will take a very long time time to stabilize with a roughly balanced number of zeros and ones.
I would like to add such generators (well, I already did but can withdraw my commit). The ones I want to add are the WELL generators (Well Equidistributed Long period Linear) created by François Panneton, Pierre L'Ecuyer and Makoto Matsumoto. They are described in their 2006 paper: Improved Long-Period Generators Based on Linear Recurrences Modulo 2, ransactions on Mathematical Software, 32, 1 (2006) which is available at [http://www.iro.umontreal.ca/~lecuyer/myftp/papers/wellrng.pdf].
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (MATH-419) adding WELL random generators
Posted by "Luc Maisonobe (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MATH-419?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Luc Maisonobe resolved MATH-419.
--------------------------------
Resolution: Fixed
Fixed in subversion repository as of r999577 for trunk (version 3.0) and r999779 for branch 2.x
> adding WELL random generators
> -----------------------------
>
> Key: MATH-419
> URL: https://issues.apache.org/jira/browse/MATH-419
> Project: Commons Math
> Issue Type: New Feature
> Affects Versions: 2.1
> Reporter: Luc Maisonobe
> Assignee: Luc Maisonobe
> Fix For: 2.2
>
>
> Up to 2.1, we have few pseudo random number generators. We have an interface RandomGenerator implemented by three classes:
> - JDKRandomGenerator that extends the JDK provided generator
> - AbstractRandomGenerator as a helper for users generators
> - BitStreamGenerator which in turn is extended only by MersenneTwister
> The JDK provided generator is a simple one that can be used only for very simple needs. The Mersenne Twister is on the other hand a fast generator with good properties well suited for Monte-Carlo simulation. It is equidistributed for generating vectors up to dimension 623 and has a huge period: 219937 - 1.
> Since Mersenne-Twister inception in 1997, some new generators have been created, retaining the good properties of Mersenne twister but removing some of its (few) drawbacks. The main one is that if initialized with a bits pool containing lots of zeroes, the pool will take a very long time time to stabilize with a roughly balanced number of zeros and ones.
> I would like to add such generators (well, I already did but can withdraw my commit). The ones I want to add are the WELL generators (Well Equidistributed Long period Linear) created by François Panneton, Pierre L'Ecuyer and Makoto Matsumoto. They are described in their 2006 paper: Improved Long-Period Generators Based on Linear Recurrences Modulo 2, ransactions on Mathematical Software, 32, 1 (2006) which is available at [http://www.iro.umontreal.ca/~lecuyer/myftp/papers/wellrng.pdf].
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Issue Comment Edited: (MATH-419) adding WELL random
generators
Posted by "Luc Maisonobe (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MATH-419?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12913208#action_12913208 ]
Luc Maisonobe edited comment on MATH-419 at 9/21/10 3:53 PM:
-------------------------------------------------------------
Fixed in subversion repository as of r999577 for trunk (version 3.0) and r999579 for branch 2.x
was (Author: luc):
Fixed in subversion repository as of r999577 for trunk (version 3.0) and r999779 for branch 2.x
> adding WELL random generators
> -----------------------------
>
> Key: MATH-419
> URL: https://issues.apache.org/jira/browse/MATH-419
> Project: Commons Math
> Issue Type: New Feature
> Affects Versions: 2.1
> Reporter: Luc Maisonobe
> Assignee: Luc Maisonobe
> Fix For: 2.2
>
>
> Up to 2.1, we have few pseudo random number generators. We have an interface RandomGenerator implemented by three classes:
> - JDKRandomGenerator that extends the JDK provided generator
> - AbstractRandomGenerator as a helper for users generators
> - BitStreamGenerator which in turn is extended only by MersenneTwister
> The JDK provided generator is a simple one that can be used only for very simple needs. The Mersenne Twister is on the other hand a fast generator with good properties well suited for Monte-Carlo simulation. It is equidistributed for generating vectors up to dimension 623 and has a huge period: 219937 - 1.
> Since Mersenne-Twister inception in 1997, some new generators have been created, retaining the good properties of Mersenne twister but removing some of its (few) drawbacks. The main one is that if initialized with a bits pool containing lots of zeroes, the pool will take a very long time time to stabilize with a roughly balanced number of zeros and ones.
> I would like to add such generators (well, I already did but can withdraw my commit). The ones I want to add are the WELL generators (Well Equidistributed Long period Linear) created by François Panneton, Pierre L'Ecuyer and Makoto Matsumoto. They are described in their 2006 paper: Improved Long-Period Generators Based on Linear Recurrences Modulo 2, ransactions on Mathematical Software, 32, 1 (2006) which is available at [http://www.iro.umontreal.ca/~lecuyer/myftp/papers/wellrng.pdf].
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.