Nektar merge requestshttps://gitlab.nektar.info/nektar/nektar/-/merge_requests2017-12-17T19:11:52Zhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/872provenance and meshing data from NekMesh2017-12-17T19:11:52ZMichael Turnerprovenance and meshing data from NekMeshThis adds provenance data to the end of an .xml mesh when written by NekMesh. It is the same as fields so gives information on git commit and time etc.
In addition to this when a mesh is made from scratch in NekMesh the meshing instruct...This adds provenance data to the end of an .xml mesh when written by NekMesh. It is the same as fields so gives information on git commit and time etc.
In addition to this when a mesh is made from scratch in NekMesh the meshing instructions are included in this metadata information to give the ability (with git commit id) to reproduce the mesh.
Future updates could add this sort of information to all input classes in NekMesh so that e.g the .xml states that it came from a .msh file and give the name and maybe md5.
Resolves #76v5.0.0Michael TurnerMichael Turnerhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/839feature/meshgraph-rework2018-01-09T15:09:44ZMichael Turnerfeature/meshgraph-reworkThis MR does a couple of things.
The primary purpose to rework meshgraph into a factory system, more consistent with the rest of nektar and alternatives can be added easily. This is to allow for the use of hdf5 meshes in the future. B...This MR does a couple of things.
The primary purpose to rework meshgraph into a factory system, more consistent with the rest of nektar and alternatives can be added easily. This is to allow for the use of hdf5 meshes in the future. By compartmentalising meshgraph with a factory this will become easier to maintain.
It has also got rid of the dimensionality of meshgraph, i.e the 1D, 2D and 3D variants, as this served no purpose other than making code duplication. The dimension of the mesh is stored within meshgraph anyway. All dimension specific codes are now "under the hood".
All IO related routines have been move to spatial domains, which is the lowest level they are needed at.
This is to allow for the removal of many meshIO methods. Previously there was a IO set for partitioning, meshgraph and NekMesh. Now there is only 1 IO routine per type (xml, hdf5). This will make mesh formats easier to maintain.
Spatialdomains has had a bit of a clean as well. Specifically the top level geometry classes have been trimmed down. It seemed that every developer that ever touched it made their own constructor and function set. This change had to be propagated through the code so there are a lot of file changes. A few small bugs have also been removed in the way periodic meshes are handled.
v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/901FieldConvert module to add a constant to a field2018-01-15T14:17:45ZSpencer SherwinFieldConvert module to add a constant to a fieldThis module allows one to add a constant to a particular field. This is useful if you want to add a fixed offset to the pressure field in the Incompressible flow solver.
e.g. to add a constant value of 1 to field 3 of infile.fld one wo...This module allows one to add a constant to a particular field. This is useful if you want to add a fixed offset to the pressure field in the Incompressible flow solver.
e.g. to add a constant value of 1 to field 3 of infile.fld one would type:
FielConvert -m addconsttofld:fieldid=3:value=1.0 infile.xml infile.fld outfile.fldv5.0.0Michael TurnerMichael Turnerhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/895WIP: Feature/interppoints pipeline2018-01-24T16:10:09ZKilian LackhoveWIP: Feature/interppoints pipelineuse existing expansion instead of loading it from a file, i.e. turn into a eConvertExpToPts module
This changes the modules behaviour: Instead of setting the values at not located points to zero, the point is discarded. See the bfs_pr...use existing expansion instead of loading it from a file, i.e. turn into a eConvertExpToPts module
This changes the modules behaviour: Instead of setting the values at not located points to zero, the point is discarded. See the bfs_probe test for an example, where only oneof the two queried points is inside the domain, so the resulting bfs_probe.dat contains only a single point instead of two with one haveing zero values.
in parallel, the module now uses the partitioning from the explist instead of partitioning the pts by itself.
Fixes #68 v5.0.0Douglas SersonDouglas Sersonhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/848Enable resetting the expansion after the MeshGraph is somehow changed2018-01-30T16:44:36ZStanislaw GepnerEnable resetting the expansion after the MeshGraph is somehow changedI am modifying the existing mesh and solution by adding / deleting elements. The MeshGraph does not offer a method to manipulate existing expansions and so new elements will not be included in the expansion. This MR adds a method to rese...I am modifying the existing mesh and solution by adding / deleting elements. The MeshGraph does not offer a method to manipulate existing expansions and so new elements will not be included in the expansion. This MR adds a method to reset expansions. It is assumed that only one type of elements is used in each expansion.v5.0.0Michael TurnerMichael Turnerhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/919[WIP] HDF5 mesh input2018-02-14T16:23:46ZDave Moxey[WIP] HDF5 mesh inputBuilding on !900, this branch reworks `MeshGraphHDF5` to provide functional HDF5 I/O of the geometry using distributed mesh partitioning. This branch depends on !899.Building on !900, this branch reworks `MeshGraphHDF5` to provide functional HDF5 I/O of the geometry using distributed mesh partitioning. This branch depends on !899.v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/554WIP: Fix deformation in parallel2018-11-29T15:38:05ZDave MoxeyWIP: Fix deformation in parallelThis MR fixes the application of the deformation field when Nektar++ is run in parallel. It additionally adds sanity checks for interior triangular face curvature to ensure consistency with edge curvature.This MR fixes the application of the deformation field when Nektar++ is run in parallel. It additionally adds sanity checks for interior triangular face curvature to ensure consistency with edge curvature.v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/653WIP: Add ability to specify initial condition from primitive variables for CFS2018-11-29T15:38:29ZDave MoxeyWIP: Add ability to specify initial condition from primitive variables for CFSThis MR adds the ability to specify an incompressible initial condition in terms of the primitive variables `u,v,w,p` into the compressible flow solver.This MR adds the ability to specify an incompressible initial condition in terms of the primitive variables `u,v,w,p` into the compressible flow solver.v5.0.0https://gitlab.nektar.info/nektar/nektar/-/merge_requests/929WIP: Feature gmres2019-02-01T11:19:03ZZhen-Guo Yanyanzhg@mail.ustc.edu.cnWIP: Feature gmresDeveloped the first version of GMRES.
Improvements are still under development for GMRES.
Please remind me if my coding have anything that is not proper.Developed the first version of GMRES.
Improvements are still under development for GMRES.
Please remind me if my coding have anything that is not proper.v5.0.0Chris CantwellChris Cantwellhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/1047WIP: Update CHANGELOG and VERSION for 5.0 release2019-08-08T08:47:25ZDave MoxeyWIP: Update CHANGELOG and VERSION for 5.0 releaseUpdate CHANGELOG and VERSION for 5.0 releaseUpdate CHANGELOG and VERSION for 5.0 releasev5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/1060WIP: Python variable coefficients and memory leak issues2019-08-20T12:48:25ZDave MoxeyWIP: Python variable coefficients and memory leak issuesThis MR addresses some issues with the Python bindings and brings a few new functionalities:
- It was possible for memory to be leaked or pre-emptively freed through the chain of `C++ allocation -> Python -> C++`, since the original C++...This MR addresses some issues with the Python bindings and brings a few new functionalities:
- It was possible for memory to be leaked or pre-emptively freed through the chain of `C++ allocation -> Python -> C++`, since the original C++ `Array` was not aware of the `Python -> C++` step. This has been addressed by changing the Python pointers to a double pointer shared by all Arrays, in a similar manner to `m_count`, so that they can be updated during the conversion step.
- Although reference counters seem to be accurate, segfaults can arise if C++ tries to deallocate Python arrays after execution has finished (i.e. whilst Python is cleaning up). Presumably this is because the gc is essentially ignoring reference counters during this step, so we bypass decrementing Python counters if Python is not initialised.
- Support for passing variable coefficients added in the `HelmSolve` call.
- A rudimentary wrapper for the `Comm` class has been added, which allows for at least a small amount of communication to be performed (mostly using `AllReduce`).v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/1056WIP: Fix/ldg implementation2019-09-13T08:56:22ZGiacomo CastiglioniWIP: Fix/ldg implementationLDG was not properly implemented, there was only a single numerical flux instead of computing left and right fluxes.
Also, the boundary condition is now implemented as u_ext = - u_int + 2*u_Dirichelet instead of u_ext = u_Dirichlet.
...LDG was not properly implemented, there was only a single numerical flux instead of computing left and right fluxes.
Also, the boundary condition is now implemented as u_ext = - u_int + 2*u_Dirichelet instead of u_ext = u_Dirichlet.
- [ ] 1D test Neumann
- [ ] LDG convergence study
- [ ] update LDGNS implementation v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/1062Various updates for the PWS2019-09-20T17:50:58ZRyan ReavetteVarious updates for the PWS- Code tidying
- Taken tube law out of hard code
- Added Viscoelasticity
- Added the Power tube law from *One-dimensional haemodynamic modeling and wave dynamics in the entire adult circulation,* Mynard (2015).- Code tidying
- Taken tube law out of hard code
- Added Viscoelasticity
- Added the Power tube law from *One-dimensional haemodynamic modeling and wave dynamics in the entire adult circulation,* Mynard (2015).v5.0.0Spencer SherwinSpencer Sherwinhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/1083WIP: Fix additional warnings from clang++-72019-10-09T13:11:14ZChris CantwellWIP: Fix additional warnings from clang++-7This MR fixes additional warnings generated by clang (and other compilers).This MR fixes additional warnings generated by clang (and other compilers).v5.0.0Chris CantwellChris Cantwellhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/1067WIP: Resolve "Arpack driver in MPI causing deadlock"2019-10-09T13:34:29ZMohammad HossainWIP: Resolve "Arpack driver in MPI causing deadlock"Closes #160Closes #160v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/833Feature/cfi_cad_engine2017-10-24T15:51:46ZMichael TurnerFeature/cfi_cad_engineThis adds a full new CAD back end to the nekmeshutils CAD system from CFI (CADfix).
There are also some small updates to the data structures of the CAD system which have been propagated through the code.
CADfix 12, which will be releas...This adds a full new CAD back end to the nekmeshutils CAD system from CFI (CADfix).
There are also some small updates to the data structures of the CAD system which have been propagated through the code.
CADfix 12, which will be released soon, will have some better facilities to build with nekmesh + cmake, for now that part is a bit of a hack but it works.v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/803feature/meshgraph-factory2017-07-07T13:26:06ZMichael Turnerfeature/meshgraph-factorySo this is the meshgraph rewrite without the HDF5 part of the system. That will come in another MR.
This now passes tests and appears to be functional.So this is the meshgraph rewrite without the HDF5 part of the system. That will come in another MR.
This now passes tests and appears to be functional.v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/794WIP: feature/meshgraph-factory2017-07-05T07:45:40ZMichael TurnerWIP: feature/meshgraph-factorySorry Dave this is a big MR.
This does a bit of a cleanup of spatial domains, specifically the way meshgraph deals with composites which was confusing and a bit convoluted.
There is a change in the meshgraph in that it is now a pro...Sorry Dave this is a big MR.
This does a bit of a cleanup of spatial domains, specifically the way meshgraph deals with composites which was confusing and a bit convoluted.
There is a change in the meshgraph in that it is now a proper factory, at this time it is one ended, i.e only XML. But this does remove the 1D 2D 3D separation which isnt needed and should make it easier to put HDF5 geometries in soon.
This MR is dependent on all the C++11 + cmake related MRs.v5.0.0Dave MoxeyDave Moxeyhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/766Require C++11 support for core LibUtilities classes2017-06-18T07:51:51ZDave MoxeyRequire C++11 support for core LibUtilities classesThis MR enables the requirement of C++11 for compiling Nektar++:
- CMake configuration is set up to require C++11
- it removes boost preprocessor headers in deference to variadic templates.
- `CommDataTypeTraits` has been refactored...This MR enables the requirement of C++11 for compiling Nektar++:
- CMake configuration is set up to require C++11
- it removes boost preprocessor headers in deference to variadic templates.
- `CommDataTypeTraits` has been refactored slightly to avoid compiler warnings with clang.
- `.clang-format` updated to C++11v5.0.0Chris CantwellChris Cantwellhttps://gitlab.nektar.info/nektar/nektar/-/merge_requests/709feature/cfi_cad_engine2017-03-02T15:54:25ZMichael Turnerfeature/cfi_cad_engineThis brings in the capability to use CFI as one of the possible CAD backends. It also has the ability to read a linear mesh from the CFI CAD file providing it has certain features.This brings in the capability to use CFI as one of the possible CAD backends. It also has the ability to read a linear mesh from the CFI CAD file providing it has certain features.v5.0.0Dave MoxeyDave Moxey