Re: my project in brief (was Re: inheritance and problems with #include statements)

[glen e. p. ropella]

I've moved this discussion to the swarm-modelling mailing list
since we've changed tracks.

At 08:25 AM 1/6/00 -0800, you wrote:
> As always, thanks for the speedy response. I can't find anything obviously 
> wrong with my interface declarations, so I'll let it be and schlog along 
> with redundant methods in my subclasses.

OK.  But, a little more code might help in debugging the problem.
Don't satisfice with ill-gotten OOD!!! [grin]

> 1) Financial Planning: I've got a Swarm model that simulates the INTERNAL 
> operations of my would-be company and generates pro-forma financial 
> statements. (It's pretty cool.)To my way of thinking, simulation provides a

VERY cool, actually.  Would you mind if I forwarded your message
to some prickly venture capitalists to help prove to them that
simulation is useful in the business domain?

Also, one of the first products your software company should
sell is an add-on library to Swarm that makes doing this kind
of thing (or entire business plan analyses) easier.

>  2) Domain Analysis: I've got a second Swarm model that simulates the 
> EXTERNAL business domain in which my would-be company aims to operate. 
> Benefits: a) it provides a great heuristic device for understanding the 
> real business dynamics in the proposed marketplace. b) it forces clear 
> articulation of the business process changes that would accompany the 
> proposed software system, and c) it allows for financial benefits 
> estimation for potential clients. (I'm still working on this part, but I 
> hope to get some cool reuse from the first effort and generate before and 
> after proformas for clients to help them with the return on investment 
> calculation. I also plan to link this model back to the first so my 
> clients' [er, their agent representations] activities drive the revenue 
> estimations in my financial models - likely to be today's task.)

This is more like the simulations I've done, except the map from
internal to a business to external (in the market) isn't well
developed in my mind.  I'd be very interested in talking in-depth
about what you've done, here.

> 3) System Testing: The software system under development is written in 
> Java. Scalability is a big issue. So is the integrity of distributed 
> transactions. I plan to let my agents 'use' the system via the Java 
> interfaces (the software system won't be able to distinguish between 
> agents and 'real' users). Of course, I won't get usability feedback (at 
> least in early versions of the model *s*), but it's a great way to get 
> some baseline system and stress testing going without bumming out real users.

This one is much fuzzier.  What relationship does Swarm have to
the Java software you're developing or the testing interface
you're building?

> BTW, I'm sort of a UML (unified modeling language) junkie. We don't talk 
> much about notational and design methods, approaches, or methodologies 
> here. Which seems a bit of a shame because the technical challenges of 
> using Swarm are the more obvious but not necessarily the more important 
> ones in building really robust and useful models. IMHO. If anyone wants to 
> talk about good object design - I'm always up for it.

All this kind of discussion belongs on the swarm-modelling list.
We used to have alot of input from Roger Burkhart and Jim Odell
about the use of things like UML in Swarm applications.  But,
they seem to have disappeared.  Jim has a couple of links on
his web page: http://www.jamesodell.com/publications
particularly:
http://www.omg.org/cgi-bin/doc?ad/99-12-01.pdf

And I have my own peculiar "systems method" that I apply
when I build simulations.  I intended to publish it before
discussing it; but, it doesn't look like I'll ever have the
time to actually write a formal paper on it.  So, I'm up
for methodology discussions, as well.

Also, some of the discussions we used to have awhile back
about developing a formalism for ABM might have renewed interest.
I've had two people ask me about it recently.  So, if anyone is
interested in pulling out their old algebra hats or their
model theory hats, we should stoke that fire again.

glen

--
glen e. p. ropella =><= Feeding the hamster wheel.  Hail Eris!
Home: http://www.trail.com/~gepr                (505) 424-0448
Work: http://www.swarm.com                      (505) 995-0818


                 ==================================
  Swarm-Modelling is for discussion of Simulation and Modelling techniques
  esp. using Swarm.  For list administration needs (esp. [un]subscribing),
  please send a message to <address@hidden> with "help" in the
  body of the message.
                 ==================================