dolfin team mailing list archive

[Johan Hake <johan.hake@xxxxxxxxx>]

On Thursday September 16 2010 14:40:12 Anders Logg wrote:
> On Thu, Sep 16, 2010 at 11:13:01PM +0200, Kristian Ølgaard wrote:
> > On 16 September 2010 22:45, Johan Hake <johan.hake@xxxxxxxxx> wrote:
> > > On Thursday September 16 2010 13:33:03 Anders Logg wrote:
> > >> On Thu, Sep 16, 2010 at 01:27:07PM -0700, Johan Hake wrote:
> > >> > On Thursday September 16 2010 13:15:51 Anders Logg wrote:
> > >> > > On Thu, Sep 16, 2010 at 01:10:55PM -0700, Johan Hake wrote:
> > >> > > > On Thursday September 16 2010 13:03:44 Anders Logg wrote:
> > >> > > > > On Thu, Sep 16, 2010 at 12:51:28PM -0700, Johan Hake wrote:
> > >> > > > > > On Thursday September 16 2010 12:27:57 Anders Logg wrote:
> > >> > > > > > > On Thu, Sep 16, 2010 at 12:23:48PM -0700, Johan Hake wrote:
> > >> > > > > > > > On Thursday September 16 2010 11:46:02 Anders Logg wrote:
> > >> > > > > > > > > On Thu, Sep 16, 2010 at 11:24:34AM -0700, Johan Hake 
wrote:
> > >> > > > > > > > > > Hello!
> > >> > > > > > > > > > 
> > >> > > > > > > > > > I added the method area to the Face class. I guess a
> > >> > > > > > > > > > Face is always a triangle so it should be safe.
> > >> > > > > > > > > 
> > >> > > > > > > > > Yes.
> > >> > > > > > > > 
> > >> > > > > > > > ok
> > >> > > > > > > > 
> > >> > > > > > > > > > There might be other methods that can be usefull
> > >> > > > > > > > > > like normal? Others?
> > >> > > > > > > > > 
> > >> > > > > > > > > Probably but can't think of any more right now.
> > >> > > > > > > > > 
> > >> > > > > > > > > > Should we do the same for Facet, but then check for
> > >> > > > > > > > > > topological dimension before making the computation,
> > >> > > > > > > > > > a la the generalized volume in CellTypes?
> > >> > > > > > > > > 
> > >> > > > > > > > > Facet should have normal but not area.
> > >> > > > > > > > 
> > >> > > > > > > > I wont add these now. I first thought they were straight
> > >> > > > > > > > forward, but all kindoff R^1, R^2, R^3 stuff needs to be
> > >> > > > > > > > checked for. Leave it for now...
> > >> > > > > > > > 
> > >> > > > > > > > Added a blueprint!
> > >> > > > > > > 
> > >> > > > > > > I thought I already did that. We have Cell.normal() which
> > >> > > > > > > should handle that.
> > >> > > > > > 
> > >> > > > > > That just calls normal of TriangleCell or TetrahedronCell.
> > >> > > > > > TriangleCell only returns a normal for topological dimension
> > >> > > > > > 2.
> > >> > > > > 
> > >> > > > > Isn't that enough? What else do you need?
> > >> > > > 
> > >> > > > For a Face I need to check if it is in R^2 or R^3.
> > >> > > > 
> > >> > > > > > It is really not difficult, it just took some more time than
> > >> > > > > > I anticipated, and I do not need the feature. I just needed
> > >> > > > > > area :)
> > >> > > > > 
> > >> > > > > Oh so you didn't need anything else. ;-)
> > >> > > > 
> > >> > > > Nope! Not for the moment.
> > >> > > > 
> > >> > > > But now it is at least straight forward to iterate over faces of
> > >> > > > a 3D mesh and calculate the area of a certain boundary domain
> > >> > > > given by a MeshFunction :)
> > >> > > 
> > >> > > Perhaps we could have
> > >> > > 
> > >> > >   double::Mesh::area(const FacetFunction& boundary_markers, uint
> > >> > >   boundary)
> > >> > > 
> > >> > > const;
> > >> > 
> > >> > +
> > >> > 
> > >> > Why not add a volume method while at it? Maybe we should let these
> > >> > be free functions as it does not always make sense to get an area
> > >> > or volume of a mesh?
> > >> 
> > >> I don't know what the dimension-independent terms are but most people
> > >> would probably accept "area" as meaning the length of the boundary of
> > >> a 2D domain.
> > > 
> > > If I had a 2D mesh and it had a method area, I would definetly think
> > > that it would return the area of that mesh. But that might be just
> > > me...
> > 
> > Anders is talking about the area of Face/Facet of a 2D mesh which is
> > where this discussion originated from.
> > 
> > Kristian
> 
> I agree it might be confusing. Perhaps measure() and
> boundary_measure() would work?

What whith:

  measure(std::string type)

which then takes a "volume", "area", or "length" argument together with an 
optional MeshFunction and subdomain?

Johan

> --
> Anders