Greetings,
I'm curious to learn about the ordering and orientation of the edge basis in Elmer. In some approaches,one sorts all edges lexicographically, and their ordering describes the basis. Does Elmer follow a similar approach?
One could also loop over elements - as given in mesh.elements - and use the local ordering as given on p.144 in the solver manual. Here, if an edge (unordered) hasn't been considered, it is added to the edge basis list. However, if the edge already exists (up to reordering), it should be omitted. Depending on the ordering of the mesh.elements one would get different bases.
Thanks in advance for your help.
Best regards,
K Batra
Order and Orientation of the Edge Basis
-
- Site Admin
- Posts: 4838
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Order and Orientation of the Edge Basis
Hi
To my understanding the orientation of the edge basis is define by the global node indexes of the nodes associated with the edge. So positive direction is from smaller to larger index (or vice versa...).
-Peter
To my understanding the orientation of the edge basis is define by the global node indexes of the nodes associated with the edge. So positive direction is from smaller to larger index (or vice versa...).
-Peter
Re: Order and Orientation of the Edge Basis
Indeed, so far so good. But does then also 1-3 come before 2-42 in the full list over edge elements? That is lexicographical?
Thanks
K Batra
Thanks
K Batra
-
- Site Admin
- Posts: 4838
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Order and Orientation of the Edge Basis
Hi
I dont think there is any ordering of the edges in that sense. To my understanding first we have elements, say tets. If edges are not needed they are not even generated. The edges probably get their edge indexes in the order they are detected. So first tet would have edges 1-4, and so on, expect when edges (=two nodes) are shared the already-used index is used. So if the last element would include nodes 1 and 3 the edge 1-3 could be even the last one identified.
I guess there could be benefit from this kind of ordering?
-Peter
I dont think there is any ordering of the edges in that sense. To my understanding first we have elements, say tets. If edges are not needed they are not even generated. The edges probably get their edge indexes in the order they are detected. So first tet would have edges 1-4, and so on, expect when edges (=two nodes) are shared the already-used index is used. So if the last element would include nodes 1 and 3 the edge 1-3 could be even the last one identified.
I guess there could be benefit from this kind of ordering?
-Peter