1
\section{Axisymetric geometries}
2
% -----------------------------
5
\cindex{axisymetric geometry}
6
\cindex{polar coordinate system}
7
\cindex{weighted functional space}
9
\fiindex{\file{.geo} mesh}
11
The coordinate system is associated to the geometry description,
12
stored together with the mesh in the \file{.geo}:
14
mkgeo_grid -t 10 -rz > square-rz.geo
17
Note the additional line in the header:
23
% The \code{geo} class also provides a way to specify the coordinates system
26
% omega.set_coordinate_system ("rz");
28
Here \code{"rz"} means that the coordinate system $(x_0,x_1)= (r,z)$.
29
Notes that the coordinate system $(z,r)$ is also supported
30
but not fully tested yet:
31
it may also be implemented by a suitable coordinate swap.
32
The \code{"cartesian"} argument string is also supported
33
and means that the usual coordinate system may be used.
34
In the \code{"rz"} case, the $L^2$ functional space is equipped with the
35
following weighted scalar product
37
(f,g) = \int_\Omega f(r,z) \, g(r,z) \, r \, dr dz
39
and all usual bilinear forms are now implemented by using this weight.
41
By this way, a program source code can handle both cartesian
42
and axisymetric systems: only the input geometry makes the
48
Thus, the coordinate system can be chosen at run time and
49
we can expect an efficient source code reduction.
51
\cindex{Poiseuille flow}
53
\code{stokes-poiseuille.cc}
55
\code{stokes-poiseuille-bubble.cc}
56
are provided in the \code{example} directory.
57
\exindex{stokes-poiseuille.cc}
58
\exindex{stokes-poiseuille-bubble.cc}
59
These examples are related to the Poiseuille flow
60
of a fluid either between parallel planes (cartesian)
61
or in a pipe (axisymetric).
62
The first example implements the Taylor-Hood approximation
63
while the second one uses the bubble-stabilized P1-P1 approximation.