Re: [Swarm-Modelling] short reading list for IBMs/ABMs in ecology?

[Doug Donalson]

Yes Alex, I'm still lurking.  I wrote my own C++/Windows GUI with lists etc.
so I am using that now (faster is better).  To answer your questions, HADES
was pre Swarm.  I will send you a couple of things I have written that may
or may not help.

Here is a paper that might be useful to you.  It compares the results of a
PDE model to those of an equivelent ABM.

Nisbet, R. M., Diehl, S., Cooper, S. D., Wilson, W. G., Donalson, D. D.,
Kratz, K., Primary productivity gradients and short-tern population dynamics
in open systems. Ecological Monographs 67: 535-553, 1997.

You might want to think about the following when approaching your orals:

The biggest argument against ABMs in Ecology has been the assumption that
adding more parameters means that the model becomes less accurate.  This is
not true.  (I call it " !(The Emperor's New Clothes) " argument). If you
pretend it is no longer there, you don't have to think aboit it.)  It comes
from the fact that as you move to simpler model classes, say from a CA based
representation of a system to an ODE based representation of the same
system, the interactions appear to become more simple and thereby more
tractable.  However, this is an illusion.  All of the mechanisms that are
explicitly represented in the CA still exist in the ODE, they are just
subsumed in the implicit structure of the model.   Over and over I have the
same discussion with eco-modelers.  "Tell me, if you think movement might be
important in this system than how can you believe that a model that has
instant teleportation of all agents to new random locations every
millisecond (the well-mixed ODE assumption) can be more accurate than a
model where you can explore the dynamical responses to different possible
spatial movement scenarios (including the well-mixed assumption) and better
understand how spatial movement might effect the overall system dynamics."
You can extend this argument to things like "logistic growth", which becomes
quite interesting when you try to implement it in a spatially explicit
model, or some of the functional responses (like the Holling Type 3
functional response.)

One of the things I am sending you is a book chapter that is in the
publication cycle at present that discusses how using different model
classes (ODE, SBD, CA, and ABM) to explore the same system can greatly
improve both the verification and validation processes in model development.
I actually have a much better (modified) version of this chapter but it is
still is prep and not at the point where it would do you any good.  It
explores how different spatial assumptions (for example grid-based vs.
continuous space) change system dynamics in a model of mussels and their
predators, sea stars.

Anyway, I'll get those out to you ASAP.  (A great philosophical question is,
"What does ASAP mean to a procrastinator?")

Cheers,

D4

PS  Have you discussed this with Cherie Briggs or is she "The Committee
Member".



----- Original Message -----
From: "Alex Lancaster" <address@hidden>
To: <address@hidden>
Cc: "Jim Anderson" <address@hidden>
Sent: Wednesday, March 26, 2003 4:28 AM
Subject: [Swarm-Modelling] short reading list for IBMs/ABMs in ecology?


> Hi folks,
>
> A question to all you agent-based [ABM] (aka individual-based [IBM])
> ecological modellers out there.
>
> I am taking my PhD qualifying exam in less than 2 months and I have a
> committee member who is a theoretical ecologist, who, while interested
> in individual-based modelling (Swarm-style), isn't an active
> practitioner (not counting "individual-oriented" stochastic models).
>
> Even though I am not an ecologist, I have chosen this person because:
>
> (1) I have worked with ecologists developing IBM/ABM models in the
> past so am familiar with many of the issues in IBM/ABM modelling (but
> not so much the "classical" theory); (2) although this faculty member
> does field work, is mainly a theory person, which is hard to find in
> biology departments, since I am completely a theory/data analysis
> person (although I have helped others on field trips!); (3) this
> ecologist seems receptive to the idea of IBMs/ABMs in ecology; (4)
> I've worked with ABMs for long enough to know I know many of the
> advantages as well as some of the standard criticisms and limitations;
> (5) I know and Swarm (and ABM tools) pretty well... ;-)
>
> My own work thus far in grad school and most of my knowledge of
> "classical", non-ABM modelling is largely in population and
> evolutionary genetics (both modelling and data analysis).  My other
> committee members consist of: a population geneticist; an evolutionary
> biologist and morphologist, and a computational cell biologist.
>
> Since it's outside of my main research focus, it's understood that I
> wouldn't be expected to know as much ecology as an ecology grad
> student, but I do need to be able intelligently address and discuss
> issues of ABM in biology, with some reference to ecology (since that's
> the speciality of this particular professor).
>
> I know that there have been various discussions and lists of papers on
> ABM/IBM in ecology on this very list (and all in the archives), but I
> have a very specific request.
>
> I am looking to put together a short (say 3 or 4 papers) reading list
> which is focused (so the questions in the exam will also be focused)
> with this committee member, given the following constraints:
>
> 1. I only know a smattering of ecology.
> 2. I have less than 2 months for reading and discussing (and I also
>    have 3 other committee members to study for)
> 3. I'd like to put my experience of Swarm and ABM in general to good
>    use (that part I know pretty well).
> 4. The papers should specifically address the usual objections raised
>    as to the need for agent-based or individual-based modelling in the
>    first place (i.e. they shouldn't assume that they are preaching to
>    the converted).
>
> I am familiar with the work of Steve Railsback, Steve Jackson, Glen
> Ropella and their colleagues at Humboldt State, and Steve has already
> kindly pointed me at some relevant papers.  I also know the work of
> Doug Donalson, in particular Donalson & Nisbet in Ecology 80:2492
> (1999).  (Incidentally, does anybody know whether Donalson's HADES
> predated Swarm, or did it use Swarm internally. Is Doug still lurking
> on this list?)
>
> Also, I know from SwarmFest's past, of Jim Anderson's work at CBR at
> U. Washington on modelling salmon, but don't which paper would be a
> good one to read in this context.
>
> I'm looking for papers that compare and constrast ABM/IBM to classical
> models (ecological or otherwise) such as Lotka-Volterra predator-prey
> dynamics, or otherwise help provide a bridge from classical
> equation-based models to ABM/IBM (e.g. setting up an ABM/IBM in a
> "special case" mode to reproduce known classical dynamics).
>
> If there's anybody out there who can recommend papers on ABM/IBM that
> would be general enough to digest, but with reference to ecology,
> please let me know.
>
> Phew, that's a long question!  Many thanks,
>
> Alex
> _______________________________________________
> Modelling mailing list
> address@hidden
> http://www.swarm.org/mailman/listinfo/modelling