Hybrid modeling?

[Stephen Guerin]

Greetings from the wireless network at the London School of Economics!
Man, it's great, but hot, here.

I'm trying to think through the issue of doing hybrid modeling. We're
working on a project that is attempting to model and visualize the
Criminal Justice System (England and Wales). Many portions of the
model only require system dynamics' macro detail while other parts
definitely require the micro detail of agent-based modeling due to the
importance of interactions (ie a particular crown court case has
particular defendents, witnesses, police officers, prosecuters, etc
that are constrained to move through the system together). It reminds
me a bit like Stuart's Buttons and Threads example where as you add
more and more connections (agent-agent constraints) in the judicial
system, it is possible that a single court delay can ripple through a
huge cluster given a critical value of connectivity.

The point of the post is to see if anyone has any good links on case
studies mixing system dynamics (process modeling) with ABM. FWIW, one
in-the-field lesson learned so far has been the confusion the client
and team can get into with hybrid models when the client is flipped
back and forth between being asked to give activity-based descriptions
for agents, and other times being asked to give more process level
descriptions for components of the system. This echoes something Eric
Bonabeau writes about in
<http://www.pnas.org/cgi/reprint/99/suppl_3/7280.pdf>

-S

____________________________________________________
http://www.redfish.com    [hidden email]
624 Agua Fria Street      office: [phone]
Santa Fe, NM 87501        mobile: [phone]

[Richard Harris-5]

Gosh,

I ask about about wireless at Warwick and they look at me like I'm from
mars! I guess I'm at a local minima.

Don't forget to include those funny wigs they have to wear in your model.
;-)

----- Original Message -----
From: "Stephen Guerin" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, August 07, 2003 6:17 PM
Subject: [FRIAM] Hybrid modeling?

> Greetings from the wireless network at the London School of Economics!
> Man, it's great, but hot, here.
>
> I'm trying to think through the issue of doing hybrid modeling. We're
> working on a project that is attempting to model and visualize the
> Criminal Justice System (England and Wales). Many portions of the
> model only require system dynamics' macro detail while other parts
> definitely require the micro detail of agent-based modeling due to the
> importance of interactions (ie a particular crown court case has
> particular defendents, witnesses, police officers, prosecuters, etc
> that are constrained to move through the system together). It reminds
> me a bit like Stuart's Buttons and Threads example where as you add
> more and more connections (agent-agent constraints) in the judicial
> system, it is possible that a single court delay can ripple through a
> huge cluster given a critical value of connectivity.
>
> The point of the post is to see if anyone has any good links on case
> studies mixing system dynamics (process modeling) with ABM. FWIW, one
> in-the-field lesson learned so far has been the confusion the client
> and team can get into with hybrid models when the client is flipped
> back and forth between being asked to give activity-based descriptions
> for agents, and other times being asked to give more process level
> descriptions for components of the system. This echoes something Eric
> Bonabeau writes about in
> <http://www.pnas.org/cgi/reprint/99/suppl_3/7280.pdf>
>
> -S
>
> ____________________________________________________
> http://www.redfish.com    [hidden email]
> 624 Agua Fria Street      office: [phone]
> Santa Fe, NM 87501        mobile: [phone]
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9AM @ Jane's Cafe
> Lecture schedule, archives, unsubscribe, etc.:
> http://www.redfish.com/friam
>
>