i plugged the two level system suggested by mr powers into stella
over the weekend, and can report that it does just fine. s and t
converge to s* and t*. i need to bring binhex home so i can send
a copy of it to mr cziko, as a modeling example.
on the subject of simultaneous equations...
there is a slight terminology difference here (as usual, it seems).
stella is perfectly capable of being set up to model solutions to
systems of equations such as those provided by mr powers. the compiler
effectively treats the existence of a level variable as a delay of one
dt interval. variables dependent on level state are calculated first,
and then the level (which is dependent on those values) is calculated
for the end of the timestep. what is disallowed (and i believe is the source
of mr cziko's error) is the following:
source/sink<-----------valve------------>level----------:
/\ :
: :
: \/
: :------------->v1
: : :
: : :
: : \/
rate<-------------v3<------------v2
where v1,v2, and v3 are mutually dependent variables which must be
calculated at the same time. in that loop there must be a delay of
some kind (preferably a level).
one thing i can suggest to anyone trying to model with stella is to
at least obtain "introduction to system dynamics modeling with dynamo",
by george richardson and alexander pugh. the stella docs do not go into
the principles very deeply, and getting a look at the underlying
assumptions of what constitutes a 'good' model is very worthwhile. even
though the book is written as a dynamo text, it is immediately applicable
to stella (since stella basically just compresses two steps of dynamo
programming into one, at the cost of some flexibility).
-----------< Cognitive Dissonance is a 20th Century Art Form >-----------
Eric Harnden (Ronin)
<HARNDEN@AUVM.BITNET> or <HARNDEN@AMERICAN.EDU>
The American University Physics Dept.
4400 Mass. Ave. NW, Washington, DC, 20016-8058
(202) 885-2748
---------------------< Join the Cognitive Dissidents >-------------------
···
from: eric harnden (920914.0900)