[Repast-developer] Re: repast in 3D

[Mark R. D. <mdi...@la...>]

Ted Phillips wrote:
> Hi Mark -- sorry to impose -- I am have emailed Phil Feldman a little 
> bit about any ideas he might have regarding a 3D-space version of Repast 
> Heatbugs and he tells me that he is working in vector space rather than 
> (heatbugs) discrete space.  I am very much a beginning programmer but am 
> intrigued by the possibility of 3D space representation of a model like 
> Heatbugs -- I realize this is computationally challenging, but I 
> wondered if you had any experience or thoughts (or advice) you might 
> share with me.   Thanks in advance.     Ted Phillips

Ted,

Hi, yes, we have been exploring 3D Spaces in discussion. I'd say that if 
your really interested in the development of 3D Spaces your welcome to 
get involved. It your actually more interested in using the spaces once 
they have been developed I'd say stick around because I think We've got 
some interesting implementations to release soon.

You shouldn't feel shy about posting this stuff to the RePast lists, 
we'll get far more interaction from interested parties that way. There 
are a number of individuals who have done some work on 3D RePast models 
including Phil and myself.

Since this is now a development project of sorts, your welcome to join 
in on the repast-developer list. You can join both lists here:
http://repast.sourceforge.net/modules.php?op=modload&name=Sections&file=index&req=viewarticle&artid=3&page=1

I am currently working with Tom Howe on the new RePast "Space" and 
"Topology" libraries. I've also created Object3DGrids and Tori for this 
library (they still need to have "neighborhooders" created for them). 
These should make 3D implementations of the Object2DGrid based models 
relatively "painless" to create.

Phil sent me his Display Class for his Vector3D space. I've been trying 
to find time to take a look at it and attempt to "visualize" the new 
Object3DGrid spaces I've written with it as they will both use 
"iteration over agent lists" for the visualization. I would also like to 
eventually see support for both the non-discrete Vector based approach 
and the discrete grid based approach.

I will attempt to get all this into the RePast CVS tree soon so others 
can have a look at it. Here is the current contents of that location:
http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/repast/repast/src/uchicago/src/sim/topology/space/

However, the directory structure here will change once I've commited the 
newer classes, it'll look more like this

.../src/sim/topology/space/d2/Object2DGrid
.../src/sim/topology/space/d3/Object3DGrid


In terms of HeatBugs, there is also the case of the underlying 
DiffuseGrid that needs be addressed. This would mean that the 
DiffuseGrid (2D) would need to get extended into a 3D implementation as 
well. It updating mechanism would need to get modified. I think this is 
entirely possible and within reach. It just requires some R&D time to 
resolve the changes to the algorthim that would be required to insert 
the 3rd dimension.

Cheers,
Mark