That was the problem. I changed my mesh file cantilever.xda to follow the
syntax in H_channel_quads.xda in adjoint_ex3 and that made it work. I
remember reading a pdf document about xda files in libmesh, can't find it
now. Could anybody explain me why my original mesh file wasn't working?
Thanks
Miguel
On Nov 18, 2015 16:40, "Miguel Angel Salazar de Troya" <
salazardetr...@gmail.com> wrote:
> I think my problem must be in my original mesh format. This simple code
> has the same problem:
>
> int main (int argc, char** argv)
>
>
>
> {
>
> // Initialize libraries.
>
> LibMeshInit init (argc, argv);
>
>
>
> // Skip adaptive examples on a non-adaptive libMesh build
>
> #ifndef LIBMESH_ENABLE_AMR
>
> libmesh_example_requires(false, "--enable-amr");
>
> #endif
>
> // Skip this default-2D example if libMesh was compiled as 1D-only.
>
> libmesh_example_requires(2 <= LIBMESH_DIM, "2D support");
>
> libMesh::out<< "Reading in and building the mesh" << std::endl;
>
> const unsigned int dim = 2;
>
> // Create a mesh, with dimension to be overridden later, on the
>
> // default MPI communicator.
>
> Mesh mesh(init.comm(),dim);
>
> // Read in the mesh
>
> mesh.read("cantilever.xda");
>
> // Create a mesh refinement object to do the initial uniform
> refinements
>
> // on the coarse grid read in from lshaped.xda
>
> MeshRefinement initial_uniform_refinements(mesh);
>
> initial_uniform_refinements.uniformly_refine(1);
>
> mesh.write("save.xda");
>
> return 0;
>
> }
>
> My mesh cantilever.xda is
>
> DEAL 003:003
> 2 # Num. Elements
> 6 # Num. Nodes
> 8 # Length of connectivity vector (Num elements * Num Nodes)
>
> 2 # Num. Boundary Conds.
> 65536 # String Size (ignore)
> 1 # Num. Element Types. We just have quads, no triangles
>
> 5 # Element types in each block. 5 Corresponds to QUAD4 in the
> libMeshEnums::ElemType namespace
> 2 # Num. of elements in each block.
> Id String
> Title String
> 0 1 4 5
> 1 2 3 4
> 0. 0. 0.
> 50. 0. 0.
> 100. 0. 0.
> 100. 50. 0.
> 50. 50. 0.
> 0. 50. 0.
> 0 3 0
> 1 2 1
> On Nov 18, 2015 15:35, "Miguel Angel Salazar de Troya" <
> salazardetr...@gmail.com> wrote:
>
>> I realized that my nodes id's are not continuous. Having 15 nodes in my
>> mesh, their id's are 0,1,2,3,4,5,6,8,10,12,14,16,18,20,22, hence why I get
>> that assertion error, which assumes that the nodes' id are within the
>> limits of the number of nodes I think. I'm going to try to get an example
>> that I can send and that reproduces the error.
>>
>> Miguel
>> On Nov 18, 2015 12:15, "Roy Stogner" <royst...@ices.utexas.edu> wrote:
>>
>>>
>>> On Wed, 18 Nov 2015, Miguel Angel Salazar de Troya wrote:
>>>
>>> I'm trying to output my mesh into a file with Mesh::write ("mesh.xda").
>>>> Because my mesh is parallel I call Mesh::allgather () first. Then I call
>>>> Mesh::write("mesh.xda") which doesn't give me problems if I don't make
>>>> any
>>>> uniform refinement, however if I do, I obtain this assertion
>>>>
>>>> Assertion `(3*local_idx+2) < coords.size()' failed.
>>>>
>>>> (3*local_idx+2) = 38
>>>>
>>>> coords.size() = 33
>>>>
>>>> At the moment of the assertion, my mesh consists of eight elements in a
>>>> 4x2
>>>> grid, coming from a 2x1 mesh before one single uniform refinement. Any
>>>> idea?
>>>>
>>>
>>> Not a clue. Can you boil this down into an example you can share?
>>>
>>> I tried to follow adjoint_ex3 and added
>>>> mesh.renumber_nodes_and_elements () before calling the write, but it
>>>> didn't
>>>> make a difference.
>>>>
>>>
>>> I'm not sure why adjoints_ex3 is doing that, but it shouldn't be
>>> necessary.
>>> ---
>>> Roy
>>>
>>
------------------------------------------------------------------------------
_______________________________________________
Libmesh-users mailing list
Libmesh-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmesh-users
