Re: Robustness Check and "A growing body of ad-hoc analysis solutions"

[Rick Riolo]

well, being someone who instigated the creation of drone,
I of course support its approach to running simulation experiments,
i.e, do that outside of the simulation system itself, since
much of its functionality can be useful across a broad range
of computational modeling and experimentation.

regarding using drone with swarm, as paul says its pretty easy 
to do, and you don't need the UM-ExpTools setup to do it.

re the UM-ExpTools:
They are NOT specialized to any particular simulation.
They ARE a hacky, prototype implementation that provides what 
I think is a set of generally useful capabilities.

As usual, because they do what I need them to do, for me
and students, and because i can teach students how to use
them in a class or two, I've just not been motivated
to re-implement the general functionality in light of all
the changes to swarm since they were first written
(ie in pre swarm-1 days!) and in light of our experiments with them.

I'd love to see someone re-implement some class that
provides this kind of functionality:
 - processing of command line arguments, in GUI or Batch mode. 
 - processing of parameters from an input-file 
 - a pre-built set of parameters, including a RNG seed, report file
      name, output directory, inputfile, and run number. 
 - easy way to add new parameters, e.g., just adding some
    simeple table entries that specify name, alias, type, etc.
 - automatic writing of parameters to the head of the report file (or
       other user opened files) in a format ready
       for re-reading, so that runs can be replicated. 
 - utility macros and methods for RNG access, parameter access, etc. 
 - hooks and helpers to write changes made via a probe into a
     parameter list into the report file
     (eg to help record what happened in a GUI-controled run)
and two or three other useful things I can't think of right now,
(or maybe have never thought of!).

- r

Rick Riolo                           address@hidden
Center for Study of Complex Systems (CSCS)
4477 Randall Lab                
University of Michigan         Ann Arbor MI 48109-1120
Phone: 734 763 3323                  Fax: 734 763 9267
http://www.pscs.umich.edu/PEOPLE/rlr-home.html

On Fri, 9 Jul 1999, Paul E. Johnson wrote:

> Date: Fri, 09 Jul 1999 09:27:13 -0500
> From: Paul E. Johnson <address@hidden>
> Reply-To: address@hidden
> To: address@hidden
> Subject: Re: Robustness Check and "A growing body of ad-hoc analysis
    solutions"
> 
> Benedikt Stefansson wrote:
> > 
> I think it makes sense
> > to provide Drone like capability in what is being presented as a basic
> > toolbox for simulation.
> > 
> 
> I've thought so all along. That's part of the reason I worked so hard to
> make RepeatingHeatbugs work, so that I could point students to a direct
> way to repeat a simulation.  
> 
> But now that I've seen how Drone can be made to work in a simple, direct
> way, I'm not sure any more.  Now that I can point students to an
> application of Drone that does not require much tedious fiddling, I am
> not so worried.  Instead of working to build this capacity into swarm,
> perhaps we should work on getting  a simple, direct explanation of how
> to setup drone (which is simple), write an Argument class object, and go
> from there.
> 
> I'm not sure how to best present this, but I may graft the simple drone
> approach onto Heatbugs and distribute it. I found it very difficult to
> make it work, but it was difficult because I was completely naive and
> untrained in argument processing and because the drone manual was
> written before Swarm had settled on an argument passing scheme and
> because the available example of drone usage relied on complicated,
> extraneous support classes that were specialized by their authors.
>  
> -- 
> Paul E. Johnson                               email: address@hidden
> Dept. of Political Science                    
> http://lark.cc.ukans.edu/~pauljohn
> University of Kansas                          Office: (785) 864-9086
> Lawrence, Kansas 66045                        FAX: (785) 864-5700
> 
>                   ==================================
>    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.