Re: built-in RNG distributions

[Rick Riolo]

Sven,
Thanks for the pointer.
I was just trying to figure out where to point new users 
who had questions  about this, e.g., in the context of 
the use of these features in the heatbugs demo.

For the hive:
I would suggest that some page or two in the regular
swarm documentation on "intro to accessing random distributions"
or some such would be a  good idea.

thanks.
- r

Rick Riolo                           address@hidden
Program for Study of Complex Systems (PSCS)
4068 Randall Lab                
University of Michigan         Ann Arbor MI 48109-1120
Phone: 313 763 3323                  Fax: 313 763 9267
http://www.pscs.umich.edu/PEOPLE/rlr-home.html

On Tue, 27 Jan 1998, Sven N. Thommesen wrote:

> Date: Tue, 27 Jan 1998 11:17:44 -0600
> From: Sven N. Thommesen <address@hidden>
> To: address@hidden
> Subject: Re: built-in RNG distributions
> 
> At 09:23 PM 1/26/1998 -0500, you wrote:
> >
> >Could someone please remind me where in the documentation
> >tree there is a description of the current 
> >built-in RNG distributions?  
> >
> >I have found one reference to them at the end very end of (a 
> >long..) swarmdocs/src/random/UsageGuide.html page.
> >But maybe there is a more accesible place they are described,
> >for beginning users (including a description of the -varySeed
> >parameter)?
> >
> >Thanks!
> > - r
> 
> Rick,
> 
> the best description of the -varySeed functionality is found in
> $SWARMHOME/src/random/docs/WHATS.NEW.in.0.75. This functionality was
> changed between random 0.7 and random 0.75. I'm sorry if a description of
> this change has not percolated into the README.Generators.v075 file, which
> is repeated in that online Usage Guide document. (I should probably get
> together with the Hive to better streamline the random documentation.)
> 
> I repeat the -varySeed comments below.
> 
> In general, my own documentation for the random library is found in the
> ../src/random/docs directory. I have no control over what the Hive puts
> into the online documentation -- but I also have no doubt that my own
> efforts could be better. Please let me know if you find info that's
> incomplete or confusing or just plain wrong ...
> 
> Cheers,
> --Sven
> 
> -----------------------------------------------------------------------
> Excerpt from ../src/random/docs/WHATS.NEW.in.0.75:
> -----------------------------------------------------------------------
> 
> 4. Generators: the macros defining 'random' seeds have
> changed, for two
> reasons: 
> 
> (a) in 0.7, if '-varySeed' was not specified on the 
> command
> line, all generators created using the
> -createWithDefault method were
> created with the *same*
> seed (DEFAULTSEED). This could lead to many
> generators
> running in lockstep, which we want to avoid.
> 
> (b) the RANDOMSEED
> macro sometimes returned 0 
> (notably, under Solaris).
> 
> The new usage is as
> follows:
> 
> A. If the program is started without '-varySeed', the
> default
> generator (randomGenerator, an MT19937 object) is started with
> the
> fixed seed DEFAULTSEED, as before.
> 
> B. If the program is started with the
> '-varySeed' switch, the
> default generator is started with seed PIDTIMESEED,
> which
> depends on the process id and calendar time and will generally
> be
> different between program runs.
> 
> C. If further generators are created using
> the -createWithDefaults
> method, then these generators are created with the
> STARTSEED seed.
> This seed value depends on the '-varySeed' flag:
>   (a) if
> '-varySeed' is set, STARTSEED invokes RANDOMSEED, which
> depends in an
> unpredictable way on program runtime. (This macro
> has changed from 0.7.)
> 
> (b) if '-varySeed' is not set, STARTSEED invokes NEXTSEED, which
> creates a
> sequence of seed values using an inline LCG and hence is
> predictable and
> repeatable.
> 
> If you want, you can of course specify '-varySeed' to get a
> different
> starting seed, but create other generators using NEXTSEED
> explicitly.
> 
> D. Generators can of course be created using seed values
> supplied 
> by the programmer or by the user/researcher.
> 
> E. The following
> macros are available for generating seed values:
>   STARTSEED (=> NEXTSEED
> or RANDOMSEED, depending on -varySeed)
>   NEXTSEED: a sequence of values
> created by an inline LCG
>   RANDOMSEED: values that depend on the program's
> runtime
> 
> F. Fixed seed values DEFAULTSEED, DEFAULTSEED1,
> DEFAULTSEED2,
> DEFAULTSEED3, DEFAULTSEED4 remain unchanged. There is also
> macro 
> FIRSTSEED, which returns the value that the default generator
> was
> started with.
> 
> --------------------------------------------------------------------
> 
>                   ==================================
>    Swarm-Support is for discussion of the technical details of the day
>    to day usage of Swarm.  For list administration needs (esp.
>    [un]subscribing), please send a message to <address@hidden>
>    with "help" in the body of the message.
>                   ==================================
> 

                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.
                  ==================================