MALI: Several Improvements related to handling Albany velocity solver #402
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…g fix for floating condition
…s to be recomputed or not The mesh needs to be recomputed non only of the vertices ice mask has changed, but also if the floating edges mask and the cell ice mask has changed. Note: in order to compare the clee mask with the previous time steps, we would need the cell mask to have 2 time levels. Howver, if we set the cell mask to have 2 time levels we get different results. This needs to be fixed.
These functions will be copied in Albany, so that the MALI can be built with shared libraries.
1. sendig now MPAS global IDs to Albany, instead of creating contiguous id
(this was created for overcoming some issues with solver. It is not needed
any more now. Old functionality can be restored by defining the
useContiguousIds preprocessor directive.)
2. Change ownership to Albany triangles (MPAS vertices). This allows to
better distribute the triangles along the interface.
A triangle is set to belong to a processor P if:
1. P owns at least 2 nodes of the triangle associated to that vertex, OR
2. all the nodes of the triangle belong to three different processors, and P owns the fortran vertex and a node OR
3. the three nodes of the triangle and the fortran vertex belong to four different processors, and P owns the triangle node with the minimum ID
It is possible to avoid changing ownership of triangles by undefining the preprocessor directive changeTrianglesOwnership.
3. Each node is owned by a processor that owns a triangle that contain the node
This avoids situations problematic to the solver
The Albany mesh will be constructed keeping the ownership of nodes and triangles as defined in MPAS Interface.
Ownership of edges is not dictated by the MPAS Interface and can be arbitrarily chosen by Albany
Cleaning
…omm lists
for vertices and cells into one function
|
@mperego , I'm finally starting to review this now. One thing I am concerned about is commit b514e40 where there still is some work to do with the time levels of the cellMask field. I propose that we remove that commit from this PR and deal with it later. That way trying to resolve that issue does not hold up merging the rest of this PR. If you are ok with that, I can modify the branch to have that commit removed, and then open a new PR with just that commit that we can finish working on later. |
|
Sure, It makes sense to me. |
|
Note: Closing and reopening as #452 in order to fix branch names and update commit history at the same time. |
|
Also opening #453 separately for item 6 in the list above. |
matthewhoffman
added a commit
that referenced
this pull request
Feb 25, 2020
Several modifications to improve MPAS/Albany interface
1. Modifications to interface to simplify the code, reduce
communications and improve the partitioning FE mesh to avoid critical
situations for the solvers:
- use MPAS global ids for FE elements entities (years ago we created
contiguous global IDs for the FE meshes to avoid some issues with
solvers that have been overcome)
- change ownership of FE triangles (Voronoi vertices) to better
distribute triangles along the interface between processors
- change ownership of FE nodes (Voronoi cells) to avoid corner
situations like having a processor that owns a node but no
triangles containing that node
2. Allow building shared libraries for Albany and Trilinos
3. Enable use of Prismatic FE elements (in alternative to tets).
4. Remove obsolete code related to LIFEV library
5. Modify writing of ascii mesh so that it can be printed in parallel
Note: updated version from #402
* matt/landice/interface_update:
Update hashes to use for Trilinos & Albany
Modify albany input files of some tests to use Prisms and remove old options
Consolidate fns to create reversed comm lists for vert. and cells into 1 fn
Remove option to send to Albany list of contiguous Ids
Several changes to the Interface_velocity_solver
Remove functions computing the tet/prism mesh connectivity
Modify write_ascii_mesh so that it can mesh can be printed in parallel.
Remove old code for LIFEV from the Interface_velocty_solver.*pp. Cleaning.
Adding boundary conditions info whe printing the ascii mesh. Minor bug fix for floating condition
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Several modifications to improve MPAS/Albany interface
Modifications to interface to simplify the code, reduce communications and improve the partitioning FE mesh to avoid critical situations for the solvers:
Allow building shared libraries for Albany and Trilinos
Enable use of Prismatic FE elements (in alternative to tets).
Remove obsolete code related to LIFEV library
Modify writing of ascii mesh so that it can be printed in parallel
Fix logic to determine whether the finite element mesh needs to be recomputed. Note: in order to compare the cell mask with the previous time steps, we would need the cell mask to have 2 time levels. However, if we set the cell mask to have 2 time levels we get different results. This needs to be fixed. At the moment the cell mask is set to have 1 time level, so it might happen that the mesh is not recomputed when it should.