Hello Stephane,
Ok, i have used the smesh instruction to generate a surface mesh showing the
effective boundary of the discretisation. That mesh shows that everything is
correct. If i understand correctly, the surface mesh is, as well as the
tetrahedrized one, obtained through the intersection of the, let's call
them, rules defining the objects of the pov file and the background mesh,
but this time only 2d triangles are used. As i said the surface mesh shows
the two "capillaries" with the correct absolute length (length is in the Z
dimension, and depends on the size of the discretisation of the Z dimension
of the background mesh that is 0.1 cm) of 50 and 52 elements, and also the
correct relative heigth.
However, when i construct the tetrahedrize one as
mesh T1 =tetrahedrize(Omega,M);
save(medit,"gcb_inic",T1);
Something strange happens. It seems that one of the capillaries shows one
element less in the z dimension and it seems that is missing from the top
since there are 0.2 cm (2 elements) between the end of the inner and outer
cylinder in the first capillary and the plot shows only one. However, the
upper end of the capillary is correctly positioned at 6.0 cm, and the
beginning starts at a distance that is shifted upwards along z one element (
0.1 cm). So it seems that it is translated upward (along z) a length of one
element. The other capillary is missing two elements this time, from the top
as well, But it is also displaced upwards a length of one element only. Why
does this happen?. The lengths of the capillaries and their position along z
are chosen so that they fit in an integer number of elements of the z
dimension,beginning with the beginning of an element and ending with the end
of an element; so the intersection should be almost exact, or not?
Moreover, is this just an effect of the construction of the visualization
mesh?, it is my problem really been solved in the more correct geometry that
can be infered from the surface mesh plot?
Can i just ignore those missing elements and displacements of the objects if
i just obtain the values of the diffusing substance with calls of the type
a=salt(x0, y0, z0)? I have tried with a discretisation along z dimension of
0.05 cm, that is, halving the length of the elements, and the results are
the same. Again the first capillary lacks one element and is shifted up one
element, while the other lacks two from the top as before and is shifted up
one element.
Thanks so much for your help Stephane
Best regards
Emilio
PS I copy here the definition of the capillaries from the pov file so that
my mail can be better understood
difference{
cylinder {
<0.195,0.3375,1.0>,<0.195,0.3375,6.0>,0.055
}
cylinder {
<0.195,0.3375,0.9>,<0.195,0.3375,5.8>,0.025
}
pigment { color rgb <1,0,0>}
}
difference{
cylinder {
<0.195,1.0125,0.8>,<0.195,1.0125,6.0>,0.055
}
cylinder {
<0.195,1.0125,0.7>,<0.195,1.0125,5.4>,0.025
}
pigment { color rgb <1,1,0>}
}
2007/1/31, Stephane Del Pino <[EMAIL PROTECTED]>:
Hello.
If I understand your problem: you think that the domain geometry is not
correctly described.
I think that the description is correct: you are using a fictitious domain
approximation. This means that the mesh does not fit the geometry. In fact
the tetrahedrization show all tetrahedra that *intersect* the domain. If
you
want to see the effective boundary of the discretization use the
instructions
mesh smesh = surface(Omega,M);
save(medit,"surface",smesh);
you will then be able to ensure that the geometry is correct.
I am sorry but I just noticed that there is a bug concerning the surface
mesh
generation from the 1.0-pre9 release, so use your 1.0-pre7 to build this
mesh.
Concerning the size of the matrix, you can use the option
solve(salt) in Omega by M
memory(matrix=none)
{
which is a bit slower but consumes much less memory.
Best regards,
Stéphane.
_______________________________________________
ff3d-users mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/ff3d-users
--
###############################################
Dr. Emilio Melero García
Laboratorio de Estudios Cristalográficos, IACT
Instituto de parasitología Lopez Neyra,
P.T. Ciencias de la Salud, E-18100 Armilla,
Granada, España
Correo elec. :emilio_arroba_lec_._csic_._es
Tfno: 958181621, lab 107
###############################################
_______________________________________________
ff3d-users mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/ff3d-users
