CGAL users discussion list

[Juan Carlos Lopez Alfonso <>]

Hello Andreas,

Thank you for your answers. 

Andreas Fabri this is the link with my set of points: http://dl.dropbox.com/u/2516160/ALLtreatmentPoints.txt

The format of my file is:

n

x1 y1  z1  m1  d1

.    .     .     .     .

.    .     .     .     .

.    .     .     .     .

xn yn  zn  mn  dn

n -> Number of Points

x, y, z -> are the coordinates of each point

m, d -> are an int and a double, which in may case represent the structure and the dose for each point. You dont need to use these information only the list of points.

Please, if you can improve the simulations in some way, please let me know!!!

How I said, I always run my simulations in Release Mode and I use the insert function (dt.insert(begin, end)). My question is about the computing time: Can I say something about this if I dont have any problem with memory (I will use a very powerful computer)? Even when my set of points are 'degenerate' (in parallel planes), Can I always expect a result of Delaunay triangulations?

Best regards and thank you again

Juan Carlos

On Thu, Nov 24, 2011 at 8:34 AM, Andreas Fabri <> wrote:

On 24/11/2011 00:26, Juan Carlos Lopez Alfonso wrote:

Hi all:

The documentation of Triangulation_3 says that constructions of Delaunay
triangulations with more than 10^7 points is compute in 87.4 seconds. In
my case, I have a set of points (3 500 000) in parallel planes and the
computing time is much greater than this (hours). On the other hand, the
same code in the same computer for a Triangulation_3 work very fast
(minutes), but for Delaunay_triangulation_3 is very slow and throw an
alloc exception (problem with memory).

Please, could anyone explain me the reasons of this problems?

Best regards and thank in advance
Juan Carlos

Hello,

The word parallel planes makes ring a bell.
If you insert the points one by one, and start with
points on plane A, and later insert those of plane B,
the first insertions of point in plane B, will
have huge conflict zones: for the first point
it are all "triangles" you have constructed so far.
Inserting all points with the insert function
that takes a range of points avoids that.

There might be an additional problem, namely that
many orientation and insphere tests will be performed
with exact arithmetic. Can you send me your data set.
I ask because we work on a solution for that.

best regards,

andreas


--
Andreas Fabri, PhD
Chief Officer, GeometryFactory
Editor, The CGAL Project

phone: +33.492.954.912    skype: andreas.fabri

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