[Trilinos-Users] problem of mapping

[Heroux, Michael A]

Antonella,

I am not quite sure what you are asking, but perhaps you might find the
FECrsMatrix class useful.  It supports common needs for finite element
computations.

Mike

On 9/15/14, 7:38 AM, "Antonella Longo" <longo at pi.ingv.it> wrote:

>I am working with an MPI finite element code, by using mapping from
>Trilinos Package, where the nodes of the mesh are distributed according
>to a map (let us call it single_mesh_nodes) with local and global
>identifier.
>
>I need to make a new map (let us call it double_mesh_nodes) which should
>include the double number of nodes. The double_mesh_nodes should be
>constructed taking as first part the single_mesh_nodes, and afterwards
>starting again from the first node but assigning it new local and global
>identifiers. 
>
>In other words, after the last global and local identifiers node
>(last_GID, and last_LID respectively) is reached, we should have again
>the first node but with a global identifier value of maximum GID +1, and
>a local identifier of maximum LID +1, and so on for the following nodes.
>
>As an example, if we had a mesh of 4 nodes, with single_mesh_nodes LIDs =
>0, 1, 2, 3, and GIDs = 0, 3, 2, 1, the double_mesh_nodes should have 4+4
>nodes with LIDs = 0,1,2,3,4,5,6,7 and GIDs = 0, 3, 2, 1, 4, 7, 6, 5, that
>corresponds to the addition of the maximum GID number of the first 4
>nodes to the single GIDs.
>
>Can you suggest me how to construct this new map?
>
>-----------------------------------------------
>Antonella Longo
>Istituto Nazionale di Geofisica e Vulcanologia
>Sezione di Pisa
>Via Uguccione della Faggiola, 32
>Pisa, Italy
>Tel. office: 050 8311939
>Cell.: 339 7532089
>
>_______________________________________________
>Trilinos-Users mailing list
>Trilinos-Users at software.sandia.gov
>https://software.sandia.gov/mailman/listinfo/trilinos-users