Path: utzoo!utgpu!news-server.csri.toronto.edu!csri.toronto.edu!corkum
Newsgroups: comp.graphics
From: corkum@csri.toronto.edu (Brent Thomas Corkum)
Subject: Re: triangulation and contouring
Message-ID: <1990Aug16.103450.19448@jarvis.csri.toronto.edu>
Keywords: triangulation
Organization: Civil Engineering, University of Toronto
References: <1990Aug15.003127.22609@NCoast.ORG> <1990Aug15.145122.13588@jarvis.csri.toronto.edu> <6123@hub.ucsb.edu>
Date: 16 Aug 90 14:34:50 GMT
Lines: 38


>>  Wow! That's two people worrying about triangulation speed.  I haven't
>> seen the Cavendish papers, my own code evolved from C.L. Lawson, in
>> "Mathematical Software III", Rice (ed.), Academic Press 1977.  Also
>> with an assist from David Shenton's thesis, Carnegie Mellon 1987, plus
>> some original stuff.  All I can say about performance is that mesh
>> operations are *trivial* compared to what awaits you when you solve
>> your finite-element system.  With adaptive mesh refinement, I hit 3000
>> triangles in no time.  Are we talking about 2D triangulations?
>> The same would be true in 3D anyway.

  I realize that compared to the actual FE calcs, triangulation is 
rather trivial but when you want to incorporate it into a user
interface and make it interactive, speed is important. In my opinion,
setting up a FE run is the most rigourous and time consuming part
of the analysis. I don't mind letting it compute the results over night
(I realize that some people run jobs that require a lot more time to
compute) but when I have to spend a day or two sitting a terminal
creating the data file(s), I get a little anxious, especially when
I find that I've made a mistake somwhere and have to repeat the entire
process.

  Anyways, enough of that, I'm lucky right now as I'm only worried
about the 2D triangulation , 3D is next but you have to crawl before
you can walk. I'm interested in locating and hearing about different
2D algorithms tah people are using for node insertion and triangulation,
and am also interested in finding out a little more about quadtree methods
for meshing. So if anyone has anything to say about there experiences
with meshing I'd be glad to here it.

  To the author of the above comment, could you elaborate on David
Shenton's thesis as to what it's about, and if it's not to much
work , give the ISBN# if there is one or who to contact for a copy.

Thanks

Brent Corkum
corkum@ecf.toronto.edu