Re: [Swarm-Modelling] beginner

[glen e. p. ropella]

Paul Johnson wrote:
> I was in this situation about 10 y ears ago.  Here's what I'd say.  If
> you are willing to invest 6 month full time effort in learning
> Unix/Linux concepts and programming in C, and probably also system
> administration skills, then you can do it.  Otherwise, you'd have to
> hire a programmer to do it for you.
> 
> I know a lot of people use MS Windows, but working in it complicates
> everything, at least for me. So get rid of it.  Install the Free

I disagree.  Windows has its peculiarities.  But, more people program
using Windows than using Linux.  So, it's pretty clear that programming
in Windows isn't that hard.  To each his own.

People with little programming expertise should avoid Swarm like the
plague.  Even if you hire someone, the number of people on the planet
who can competently install and use Swarm is quite small and getting
smaller by the minute.  Plus, Swarm models are fragile in proportion to
the competency of the programmer.  A better place to start might be
something like Repast <http://repast.sourceforge.net>, perhaps the
Python layer (though you'll find more and better examples in Java).

If one is willing to switch operating systems, then Ubuntu is very
simple to install and begin using; so, I'd recommend that.

I agree about the IDEs.  However, I've come to find both Eclipse and
Netbeans useful at times, especially given the sheer volume of classes
and functions available in Java, let alone Repast.

If, however, you want to learn _modeling_ thoroughly, then I recommend
either Mason <http://cs.gmu.edu/~eclab/projects/mason/>, SimPy
<simpy.sourceforge.net>, or SIM_AGENT
<http://www.cs.bham.ac.uk/research/projects/poplog/packages/simagent.html>.
 These packages are much simpler than Repast or Swarm and force you to
think hard about things like scheduling, which is critical to modeling
and simulation.  Of them Mason is the more Swarm-like but the others are
equally useful.

>From another perspective, if you intend to learn programming to do this,
then Python is a good language to learn because it's useful in other
contexts and is object-oriented and interpreted (rather than compiled).
 I'd recommend a simulation tool that uses Python over something like
NetLogo <http://ccl.northwestern.edu/netlogo/> for that reason.  While
NetLogo is easy to use and understand, it is difficult to use for
complicated models for various technical reasons.  And although learning
Logo will help you understand programming, in general, it's not used for
many other things.

Given all Hilit's criteria, I think Repast is the best option and it
works fine on Windows.

-- 
glen e. p. ropella, 971-219-3846, http://tempusdictum.com
Rock journalism is people who can't write interviewing people who can't
talk for people who can't read. -- Frank Zappa