LISTSERV mailing list manager LISTSERV 16.0

Help for ECJ-INTEREST-L Archives


ECJ-INTEREST-L Archives

ECJ-INTEREST-L Archives


ECJ-INTEREST-L@LISTSERV.GMU.EDU


View:

Message:

[

First

|

Previous

|

Next

|

Last

]

By Topic:

[

First

|

Previous

|

Next

|

Last

]

By Author:

[

First

|

Previous

|

Next

|

Last

]

Font:

Proportional Font

LISTSERV Archives

LISTSERV Archives

ECJ-INTEREST-L Home

ECJ-INTEREST-L Home

ECJ-INTEREST-L  July 2005

ECJ-INTEREST-L July 2005

Subject:

Re: How do you add to statistics files?

From:

Sean Luke <[log in to unmask]>

Reply-To:

ECJ Evolutionary Computation Toolkit <[log in to unmask]>

Date:

Wed, 20 Jul 2005 11:55:48 -0400

Content-Type:

TEXT/PLAIN

Parts/Attachments:

Parts/Attachments

TEXT/PLAIN (97 lines)

I actually had a similar question and thought that it was as little
more straight forward than this. Generally, for statistical
significance, you want to run an experiment/solve a problem many
times, each time with a different random number seed, and have some
detailed statistics on each run while having an aggregation of
experimental data such as number of evaluations and whether or not
the ideal individual was found.

Is there a hook to re-seed the random number generator?

Steve, I'm not positive what you're referring to by re-seeding the RNG,
but the MersenneTwister classes can be reseeded by calling their setSeed
methods.

But let me delve in a little bit about how to make scripts and how the RNG
works in ECJ.  The random number generator's seed parameters, one per
thread, are seed.0, seed.1, etc., and they are set either to the number
zero, or to the string 'time' -- in which the wall clock time is used as a
seed, as in

        seed.0 = 31234
or
        seed.0 = time

If multiple seeds use 'time', then the time values are incremented, so you
don't have to worry that these two will have the same values:

        seed.0 = time
        seed.1 = time

The other common need is to write different stat files.  The standard stat
filename parameter is defined in stat.file

        stat.file = $out.stat

The '$' means 'in the same directory as where you ran the program' (as
opposed to 'in the same directory as the params file').

For years we've typically run ECJ using a UNIX shell script.  Imagine that
you have a statistics output which generates the final result (which you
care about) on a single line at the end of the run -- this is commonly the
case for us using, say, KozaShortStatistics.  Now let's say you want to
run 10 random jobs each of 7 different tournament selection sizes, then
you might issue a script like this:

#! /bin/tcsh
set arbitrary_offset=200031
foreach tournament (1 2 3 4 5 6 7)
        touch tourn.${tournament}.out
        rm tourn.${tournament}.out
        set job_max=10
        foreach job (0 1 2 3 4 5 6 7 8 9)
                @ seed = $tournament * $job_max + $job + $arbitrary_offset
                ec.Evolve -file my.params \
                        -p seed.0=$seed \
                        -p select.tournament.size = $tournament
                tail -1 out.stat >> tourn.${tournament}.out
        end
end
rm out.stat

And there you go.  7 files each called tourn.1.out, tourn.2.out, etc.,
with the final results of 10 jobs, one per line, in the files.  Different
seeds for every run.  If you wanted to save out different statistics
files, you could have added:

                        -p stat.file=out.${tournament}.${job}.stat \

If you wanted different parameter files, maybe one per tournament, called
1.params, 2.params, etc., you could say:

                ec.Evolve -file ${tournament}.params \

You could conceivably also do it in a scripting language like BeanShell
(btw, if you've not played with BeanShell, you really should do so
immediately -- www.beanshell.org).  BeanShell is basically a command-line
for java.

Realizing that not everyone runs in a Linux or MacOS X environment, we
have also added some rudimentary job facilities to ECJ.  The simplest
approach is to run ECJ with the jobs parameter:

        java ec.Evolve -file my.params -p jobs=10 -p seed.0=0

This will run ECJ 10 times, using seeds 0 through 10 respectively.  Each
job's statistics file will be written out separately as job.N.out.stat
where N is the job number.  Try it!

The jobs parameter reflects our basic default approach to handling jobs in
java code directly.  Want to do something more advanced?  If you take a
peek into the Evolve java code, you'll notice that we've rearranged it
considerably.  In particular, the main() function has been rewritten in
such a way as to make it possible for you to do all sorts of job loops,
loops within loops, whatever you need, if you so desire to do so in java
code rather than an outer script.

Sean

Top of Message | Previous Page | Permalink

Advanced Options


Options

Log In

Log In

Get Password

Get Password


Search Archives

Search Archives


Subscribe or Unsubscribe

Subscribe or Unsubscribe


Archives

September 2019
August 2019
June 2019
April 2019
March 2019
January 2019
December 2018
November 2018
October 2018
July 2018
May 2018
January 2018
December 2017
November 2017
October 2017
September 2017
August 2017
July 2017
June 2017
May 2017
April 2017
March 2017
February 2017
January 2017
December 2016
November 2016
October 2016
September 2016
August 2016
July 2016
June 2016
May 2016
April 2016
March 2016
February 2016
January 2016
December 2015
November 2015
October 2015
September 2015
August 2015
July 2015
June 2015
May 2015
April 2015
March 2015
February 2015
January 2015
December 2014
November 2014
September 2014
August 2014
July 2014
June 2014
May 2014
April 2014
March 2014
February 2014
January 2014
December 2013
November 2013
October 2013
September 2013
August 2013
July 2013
June 2013
May 2013
April 2013
March 2013
February 2013
January 2013
December 2012
November 2012
October 2012
September 2012
July 2012
June 2012
May 2012
April 2012
March 2012
February 2012
January 2012
December 2011
November 2011
October 2011
September 2011
August 2011
July 2011
June 2011
May 2011
April 2011
March 2011
February 2011
January 2011
December 2010
November 2010
October 2010
September 2010
August 2010
July 2010
June 2010
May 2010
April 2010
March 2010
February 2010
January 2010
December 2009
November 2009
October 2009
September 2009
August 2009
July 2009
June 2009
May 2009
April 2009
March 2009
February 2009
January 2009
December 2008
November 2008
October 2008
September 2008
August 2008
July 2008
June 2008
May 2008
April 2008
March 2008
February 2008
January 2008
December 2007
November 2007
October 2007
September 2007
August 2007
July 2007
June 2007
May 2007
April 2007
March 2007
February 2007
January 2007
December 2006
November 2006
October 2006
September 2006
August 2006
July 2006
June 2006
May 2006
April 2006
March 2006
February 2006
January 2006
December 2005
November 2005
October 2005
September 2005
August 2005
July 2005
June 2005
May 2005
April 2005
March 2005
February 2005
January 2005
December 2004
November 2004
September 2004
August 2004
July 2004
June 2004
May 2004
April 2004
March 2004

ATOM RSS1 RSS2



LISTSERV.GMU.EDU

CataList Email List Search Powered by the LISTSERV Email List Manager