Dear all,

I agree with Titto's idea:
>
> > Something that might be used to define in a formal and unambiguous
> > way the main characteristics of a simulation (kinds of agents,
> > employed algorithms, environment definition, update policy,. etc.)
> > while still being easily readable and understandable for a (trained)
> > human.

in that a simulation model specification language should be both, formal and  
unambiguous ...
>
> I would like to give an other characterisation of the medium level that
> might help and a slightly different scenario how it might be used:
>
> (1) The medium level is not necessarily a language, it could also be a
> graphical notation or a computer mediated datastructure.

These are both languages, too, though not in the ordinary sense.

I think a good specification language should, at the same time, be  
executable (and it should have a graphical representation, but not all  
aspects of a simulation model can be understood from a mere graphical  
representation). Ten years ago, we designed a language which covers a large  
span of social simulation, namely MIMOSE, which freed users from programming  
all the technical overhead over and over again, and we are now starting to  
develop a new toolbox --- see  
http://www.uni-koblenz.de/~moeh/projekte/framework.html (of which an updated  
version will be available soon, the one published there was what came out at  
long discussions during the 1997 Dagstuhl seminar on social science  
microsimulation.
>
> (2) The deployment is not forced to be formal or unambiguous. This
> is because the medium level will relate to natural language.

If a model specification language were executable, than we wouldn't even  
need the low level programming any more!

I am still convinced that the medium level should be formal and unambiguous  
(otherwise we can't use it directly for simulation replication, and if it is  
as ambiguous as natural language, then often natural language is better  
understandable than even UML). UML would give (in so far I agree) a good  
overview of, for instance, which types of agents make up a model, but it is  
not so easy to describe in languages like UML how these agents will behave.  
And if I want to replicate a model published in a journal, it is this which I  
must know. So what we need is something like an object-oriented language  
which allows several stages of refinement where on every next stage of  
refinement the model description becomes less and less ambiguous. More on  
this in a research proposal we are just about to write, and at the first SIG  
ABSS meeting in April.

Regards

Klaus

---------------------------------
Prof. Dr. Klaus G. Troitzsch --- Universitaet Koblenz-Landau                 
Tel.: +49-261-287-2643
      +49-261-33762 (home)
Fax:  +49-261-287-2642
e-mail: [log in to unmask]
WWW: http://www.uni-koblenz.de/~kgt



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%