CGAL users discussion list

["Grzegorz Hermanowicz" <>]

Thanks,
I'll take closer look at the papers. As for the precise needs: This is still work in progress. I have only one semester of mostly theoretical computational geometry course. Basic algorithms, constructions and approaches. It seems I'm still stuck in ~80'es.  So right now I'm in the process of bringing myself up to date with current work and tools.    

As for the general idea: 
I have a population of individuals placed in a closed 2d space(inside of polygon, torus, sphere). Each individual is described by its position in the given space, his age and genome, and maybe some additional feauters. Individuals can change their location to another valid location,  look for mate(s) in vicinity and try to procreate if they find one. Individuals can die to overpopulation(global, local), genetic defects, random events, and so on. All depends on the biological question asked. Most of the questions circulate around finding interaction between evolutionary strategies, genome structure and external factors like mating strategies, environmental factors and so on. I'm not biologist so I can't really bring it closer to you. If you are interested take closer look at this page http://www.smorfland.uni.wroc.pl/index.php?n=Main.HomePage
I'm trying to design most general simulation model that would not need a total rewrite for every proposed modification. I may sacrifice some additional computing time to save on the programing time. But of course O(n^2) isn't  what I hope for :)   

From previous work of my team we know that those simulations can be pretty sensitive to some innocent asumptions.
For example if we put our population onto closed non-periodic space. Individuals at the borders will have different probability of finding a mate than the ones being in the middle. Making space periodic solves this problem. 

Other exemplary problem that can arise is finiding exclusive mates. In this case if a pair of individuals choses to have  offspring they should  not be a valid partners for other individuals in a given turn. If programed badly it may lead to situation where individuals processed first have greater chance of finding mates than ones being processed last.  Not realy a CG problem though, unless we define moving individuals as a segments and mating oportunity as segment intersection. Now we must "only" find all intersections that are "first" for both intersecting segments.  

There is quite few tricky areas in what would seem and I hope is a simple problem.     

Hope that cleared things a little. 

Cheers 
    Grzegorz Hermanowicz

2008/11/5 Monique Teillaud <>

Hi Grzegorz,

I will let people who know more than me about the CGAL spatial search packages answer your first questions.

I will focus on the last one:

I have another, I hope easy question. I would like to put my simulation
on the closed 2d torus like surface. Does CGAL allow that? At the
moment I'm just curious is it something fairly easy to achieve or thing
that would need a CGAL developers skills to do?

Manuel Caroli and I have some work in progress, and a prototype running, for the 3D case, you can have a look at
http://www-sop.inria.fr/members/Monique.Teillaud/other-geometries/
for links to (preliminary) papers.
The 2d would not be harder, of course, but we are not really planning to work on the implementation, due to lack of time.

You might have been interested in the past worshop in periodic spaces
http://www.cgal.org/Events/PeriodicSpacesWorkshop/
but it took place 2 weeks ago already.

Still, I would be interested in knowing your application and your needs more precisely.

best
       Monique Teillaud
--

You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://lists-sop.inria.fr/wws/info/cgal-discuss