GitLab

Fix/GetBoundaryToElemtMap3 dh1 d consistency

This MR changes the ordering from DisContField3DHomogeneous1D::GetBoundaryToElmtMap to make it consistent with m_bndCondExpansions (i.e. sorting by boundary, instead of sorting by plane). This allows the boundary expansion from 3DH1D to be treated in the same way as the non-homogeneous case, reducing the need for expansion-specific code.

All parts of the code assuming the old ordering were updated to take these changes into account. 

See merge request !526

Fix bug in ExpList1D

This fixes a bug in ExpList1D which may lead to invalid .vtu files.

See merge request !531

Fix bug in ExpListHomogeneous2D

This fixes a bug which prevents FieldConvert from working with 3DH2D field files.

See merge request !529

Fix bug in ExpList2DH1D

This fixes a bug in ExpList2DH1D, caused by m_coeffs and m_phys not being initialised properly. This can lead to unpredictable behaviour, for example when extracting data from a field file with fewer planes.

See merge request !528

Make DisContField3DHomogeneous1D::GetBoundaryToElmtMap consistent with the boundary condition expansions ordering

Fix/extractsurface3DH1D

This MR extends the extract surface module of FieldConvert to work with 3DH1D expansions. The addnormals option is also extended to 2D and 3DH1D expansions.

I had to force FieldConvert to declare all expansions using NewField=true, since otherwise it would not work. This seems to be a problem with the copy constructor from ContField3DH1D (maybe with the boundary expansions?).

See merge request !521

Fix bug that affects using the range option with variableP

This fixes a bug in ExtractDataToCoeffs, which prevents the range option of FieldConvert from working with variable order expansions.

See merge request !522

feature/mc-linearise

I got a little fed-up with having to open the xml and remove the <Curved> tag section to linearise the mesh (a step needed for the linear elastic solver). Doing this manually was a nightmare with large meshes.

This module for meshconvert simply goes through all the edges and faces and removes the high-order nodes.

See merge request !509

Fix issue with HexGeom::v_GetDir that affects hexahedra defined with heterogeneous basis functions

This MR fixes an issue with `HexGeom::v_GetDir` that affects hexahedra defined with heterogeneous basis functions.

See merge request !520

fix/mc-face-curvature

The face curvature of elements was not being considered by meshconvert in relation to the spatialdomains element type. Therefore when considering the validity of elements in terms of positive jacobian, meshconvert was identifying large numbers of element which were invalid as valid. This has now been fixed.

I have tested this against other parts of nektar (fieldconvert/solvers) and everything now agrees on which elements are invalid or not.

Also a small amount of tab clear up.

See merge request !511

Fix bug with relative edge/face orientation, downgrade hex test to order 5 (Gmsh hex output seems to break at orders >= 6)

FilterHistoryPoints: fix insufficient field width of time, add spacing to point …

…locations and port to boost::format

See merge request !518

Feature/variable p

This MR implements modifications in the AssemblyMap to allow the use of variable order in 2D simulations using CG.

See merge request !497

FilterHistoryPoints: fix insufficient field width of time, add spacing to point locations and port to boost::format

small bug fixes to interpolate points routine in fieldconvert



See merge request !515

Fix/ape restart



See merge request !517

Feature/BCs

Modified dirichlet boundary conditions set-up to read both analytic expression and condition from file and multiply together. 

See merge request !514

fix/evaluate once

ensure we only load fields from disk once.

See merge request !500