Hello Sean, Tony, Claudio, Mike,
thank you all for your answers, community = top notch :-))
* Claudio wrote:
> you should consider MASON if you need:
>-computational and visualization separation for
>-lots of runs (>103) without viz every run ;-)
>-checkpointing for stats analysis
>-evolutionary computational methods, as common for assessing learning.
That is indeed exactly what I need. Especially visualization is very
important, as I want to experiment with a few different approaches here.
>We will make sure to send you the revised
>paper as we develop it.
Please do, I would be very glad, I've already printed the paper and am
currently reading it :-)
>Both Epstein and Axtell
>(now at our Center, no longer Brookings) provided input.
Wow. Cool! :-) I thoroughly enjoyed their Sugarscape book and it was
a source of great inspiration.
* Tony wrote
>I'm just about to release my source code (refreshing it for MASON 11
>and doing general cleanup/commenting) and you'll be welcome to look at
>or use it. I implemented 70-75% of the rules and outcomes in Growing
I'm looking forward to your code.
>Any kind of complex cognitive architecture will fly against the
>philosophy of Sugarscape, but is interesting that the agents have no
>overt memory to use in reasoning/(ir)rational decision making.
That is the big question - of course, the traditional CAS approach is
to keep the individual components simple. But in the social simulation
community voices have been raised (and I tend to concur - with reservations ;-) -
that to make sensible descriptions of human societies one cannot completely
neglect the cognitive aspect.
I think one should iterete into more complex agents slowly, and see what happens.
Generative science in the best sense ;-)
> As agents/models become more complex they will
>become more difficult to understand and in turn establishing causality
>between code and simulation outcomes. Given the very little
>replication that has occurred in this field, in general, there are many
That is quite true. I am just starting out with my Ph.D on this topic
(cultural diffusion modeling), and I'm not yet sure where it will take me -
more into the philosophy of science section, as in "how applicable are the
results", "is generative science real science" etc; or into the design of
cognitive agents useful for simulation; or a return to the simple agent and
adjusting the interplay of agents. All exciting questions :-)
One more thing: what I certainly intend to do is distribute the simulation
in a second step (long term goal); if I use cognitive agents, there will be no
other way to do simulation sensibly - does MASON lend itself easily to this?
I know that you can add distributed support to Repast Models quite easily.
Kind Regards and thank you for your responses,