Re: Numerical integrals on CAD models




In article <030c9a84-5d16-4c31-bd68-461331d03922@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
giacomo.ciani@xxxxxxxxx writes:
Hi all,

I'm not sure I'm wrinting in the right newsgroup, but it seems the
right place where to find people that can give me suggestions about my
problem.

What I'm looking for is a (simple?) program that can import a CAD
geometry (in a common format), mesh it and compute volume integrals of
arbitrary, analytical functions on that geometry (a stupid example,
integrate the density to find the center of mass).

No integral/differential equation, no time evolution. Just a numerical
integral. The only thing I need, becouse I'm working with rather
complicate geometries, is that it can import them istead of requiring
to set the boundaries of integration by hand...

Are these kind of tasks usually performed with complex FEM softwares
(that is like shooting to a fly with a cannon, I think)? Anyone knows
a rather simple program (if free, it would be great) that can do what
I need?

Thanks

Giacomo

since I don't know how your CAD data look like, maybe this is
of no help for you, but
cubpack
in
http://www.netlib.org/toms
has the possibility to integrate functions of two variables over
rather general 2D domains and is very flexible in tis application.

if you describe the domain by a set of vertices then also use
of delaunay triangulation, high order smooth interpolation on it
plus exact integration of the piecewise polynomial function might
be of help,
software in several parts in the same directory
hth
peter



hth
peter

.



Relevant Pages

  • Re: Why not a setf-able NIL ?
    ... Just suppose you want to repeat calculations for a non-euclidean space. ... It is independent of geometry, ... physics, coordinate system, or anything but the most basic assumptions ... your probability and integration and all sorts of other ...
    (comp.lang.lisp)
  • Re: EndPoints in Integration
    ... >spellucci@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx (Peter Spellucci) writes: ... >>>I have a naive question on the handling of divergence of endpoints by ... >>>If the integration is divergent over both endpoints of its integration ...
    (sci.math.num-analysis)
  • Re: Definite integral with huge variations in magnitude
    ... quad (adaptive simpson) or quadl tends to make me ... Thanks Peter. ... already done that with the Romberg and AdaptQuad integration ...
    (sci.math.num-analysis)
  • Re: EndPoints in Integration
    ... >spellucci@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx (Peter Spellucci) writes: ... >>>I have a naive question on the handling of divergence of endpoints by ... >>>If the integration is divergent over both endpoints of its integration ...
    (sci.math.num-analysis)
  • Re: EndPoints in Integration
    ... >spellucci@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx (Peter Spellucci) writes: ... >>>I have a naive question on the handling of divergence of endpoints by ... >>>If the integration is divergent over both endpoints of its integration ...
    (sci.math.num-analysis)