LGR Egrid #4354
LGR Egrid #4354
Conversation
…ment of the save method for the LGR EclipseGrid - EclipseGrid object and children are initialized from .EGRID file generated on standard simulator
EGrid.hpp seems to contain a bug where get<float>(COORD) cannot catpure the global COORD in EGRIDs, EclipseGrid has been enhaced with functions to output .EGRIDS for LGR - unfinished work - devising a mechanism to define HOSTCE
arturcastiel
force-pushed
the
lgr-egrid
branch
from
80c98cb to
197ab9a
Compare
|
@bska I believe I have solved the merged conflicts. |
|
jenkins build this please |
|
jenkins build this please |
…at register it was mistakenly removed
|
jenkins build this please |
|
jenkins build this please |
There was a problem hiding this comment.
The size of this PR means that it's not really possible to do a full review of the details and its algorithms. I expect that there will be follow-up work to simplify this code. That said, it's starting to look good. With the possible exceptions of a vector that's still passed by value, a couple of parameters that are unused and an unused local variable, this is starting to look ready for merging to me.
I will merge the local branch I am working right now that divides the huge function into smaller parts. I tried to achieve here a smaller PR. |
…tions used in EclipseGrid::init_children_host_cells
…tions used in EclipseGrid::init_children_host_cells
|
I believe I have solved the merging and rebasing conflicts and everything is ready to be merged. |
|
jenkins build this please |
prior to OpenMP 3.0, unsigned loop variables is unsupported.
…ment of the save method for the LGR EclipseGrid - EclipseGrid object and children are initialized from .EGRID file generated on standard simulator
EGrid.hpp seems to contain a bug where get<float>(COORD) cannot catpure the global COORD in EGRIDs, EclipseGrid has been enhaced with functions to output .EGRIDS for LGR - unfinished work - devising a mechanism to define HOSTCE
…tions used in EclipseGrid::init_children_host_cells
|
I have redone the rebase, properly, this time. |
|
jenkins build this please |
LGR Egrid
This is an ongoing effort to create the visualization routines for LGR. This PR specifically tackles the extension of the EGRID output files
LgrOutputTest- new tests routines to ensure the EGRID files are consistent with the standard simulator counterpart. The process of retrieving the refinement will take place in opm-grid, therefore we have added routines and tests to this test that simulate obtaining the COORD and ZCORN from the OUTPUT of the standard simulator. OPM initializesEclipseGridobject with this and output the EGRID. After that some fields on the file are tested.EclipseGridinialization for LGR consists:save_all_lgr_labelscreate_lgr_cells_treeinitializeLGRObjectIndicespropagateParentIndicesToLGRChildreninitializeLGRTreeIndicesparseGlobalReferenceToChildrenEclipseGrid::init_children_host_cells- maps which refined cells(LGR refined cell) belong to host cells (coarse mesh cells refined). The current approach for solving this uses an algorithm that checks if a point is contained inside a given polyhedron. This is done by using the test point to divide the polyhedron into multiply tetrahedron. If the volume of the sum of these tetrahedron is equal to the volume of the original volume, the point is internal to the polyhedron. Note: Boundary boxes are used to avoid to a precheck and avoid unnecessary computation. Although this implementation is not optimal, it is robust, for general unstructured grids. Next PR will contain a faster algorithm, and this implementation will become a fallback method.