[Libmesh-users] Order of points of a side

[John P. <pet...@cf...>]

Hi Tim,

I believe the clockwise ordering is the one you will get.  See, for
example, cell_hex8.C, Hex8::side_nodes_map.  For side 0 we have the
array {0,3,2,1}.  According to the ASCII art picture in cell_hex8.h,
we can see that this corresponds to the clockwise ordering of nodes
on the bottom face of the hex.  Also, we use these sides to perform
boundary integrations, so they must be well-formed from that
perspective as well.

I sense a leading question though, have you already found some behavior
which contradicts this?

-John


Tim Kr,bv(Bger writes:
 > Dear all,
 > 
 > In a certain situation, I want to traverse all vertices of a certain 
 > side of a three-dimensional element.  I need only the coordinates of 
 > these vertices.  My idea is this:
 > 
 > AutoPtr<Elem> side = elem->build_side(num);
 > for(l=0; l<side->n_nodes(); l++)
 >    {
 >      if(side->is_vertex(l))
 >        {
 >          do_something(side->point(l));
 >        }
 >    }
 > 
 > However, it is important for me that I get the points in a sensible 
 > order:
 > 
 > A--------B
 > |        |
 > |        |
 > |        |
 > |        |
 > D--------C
 > 
 > In this situation, I need something like (A,B,C,D) or (C,B,A,D), but 
 > not (A,C,B,D).  I hope you understand what I mean; I currently can't 
 > think of the correct English word.
 > 
 > Is it guaranteed by the above code that I get a suitable order of the 
 > vertices?  If not, what should I do?
 > 
 > (Currently, I'm working with tetrahedrons, so that the sides are 
 > triangles, and thus *any* order is suitable for me; but I want 
 > everything to work with hexes as well.)
 > 
 > Best Regards,
 > 
 > Tim