Gmsh is great !

Bojan Niceno niceno at
Mon Oct 5 10:46:35 CEST 1998

Dear Jean-Francois & Christophe,

First of all, thank you for putting such a usefull piece of 
code on the Web. It is so usefull and easy-to-use. It took
me 20 minutes to get acquinted with it to the level at which I can
make my own grids. Several things you have in your code done 
are geniuos. I mean: input file format, simultaneous editing
of the input file and inserting new nodes from the window,
representation of closed surfaces and volumes ... all in
all GREAT !  You make all the commercial grid generators
bust the dust. 

Ok, now with the questions:

1. I could understand only the .msh output file format, the
   other two .unv and .grsh I could not understand at all.
   Is it possible to get some further explanations on them ?

2. It seem to me that you can improve the meshing of the 
   surfaces. I mean, they are not bad, but sometimes the
   situations like these occur:

    |\     /|
    | \   / |
    |  \ /  |
    |   O   |
    |  / \  |
    | /   \ |
    |/     \|


    / \  |  / \
   /   \ | /   \
  /     \|/     \
  \     /|\     /
   \   / | \   / 
    \ /  |  \ /

    I want to say that nodes with less then five and more then
    seven surrounding triangles are undesirable. And yet, there
    is a very simple technique to avoid such thing. Maybe you
    know it allready, but is called "mesh relaxation" and is
    described in:

   W. H. Frey and D. A. Field, "Mesh Relaxation: A New Technique 
   for Improving Triangulations", Int. J. Numer. Meth. Eng. 
   31, 1121-1133 (1991)

    It is quite simple, and yet very effective.
    Maybe you plan to implement it allready ?

3. Where are the versions for HP-UX and Linux. There is only 
   something for AIX, but I don't have a clue what it is ?
4. Is it possible to get a source of the code ? If you don't 
   have the time to implement the above mentioned mesh relax-
   ation technique, I am willing to try to implement it. I have
   some experience with mesh generation (see for instance:
   I would also like to implement two new output file for-
   mats. If you don't like to give the full source, you can
   even consider the possibility to give just one procedure
   from which one has acces to data on nodes, elements and sides
   and which could be later linked to your code. I am not sure
   if I was clear enough, but I think you could only benefit
   from it. It could be also very usefull to make for new
   output file formats.

I think that was enough for the first time.

	With very kind regards


Bojan Niceno                  Delft University of Technology
Section Thermofluids          Lorentzweg 1
Applied Sciences              2628 CJ Delft 
Phone 015-2786125             The Netherlands
Fax   015-2781204