CHANGELOG.md

Changelog

v5.7.0

Library

• Modified MatrixFreeOp library switch initialisation to use BOOST_PP (!1794)
• Fix memory-leak with LowEnergyBlock preconditioner for time-updated matrices (!1627)
• Fix Fourier expansion integration weights are related test (!1803)
• Introduced the MatrixFree implementation for the PhysInterp1DScaled operator and tidying up MatrixFreeOps (!1812)
• Separate MeshGraph input/output functions into a new class (!1778)
• Added checkpoint file writing start time in the fieldconvert filter (!1789)
• Fix fieldconvert filter incorrect boundary values (!1789)
• Fix numerical precision issues with filters OutputStartTime (!1789)
• Fix AdaptiveSFD for MPI (!1821)
• Fix IterativeStaticCond when using absolute tolerance (!1850)
• Fix deadlock by scotch with multi-threading support (!1853)
• Fixed L2norm for FilterError (!1871)

IncNavierStokesSolver

• Fix initial and boundary conditions in the moving reference frame (!1692, !1820)
• Fix memory-leak for the Mixed_CG_Discontinuous projection when initializing the traceMep (!1806)
• Add synthetic turbulence generation for the incompressible solver (!1664)
• Fix a uninitialized parameter in VCS (!1880)

ShallowWaterSolver

• Implement implicit time-discritization (!1784)

NekMesh

• Added revolve module (!1825)

FieldConvert

• Add vortexinducedvelocity module to compute the vortex-induced velocity (!1824)
• Add a module to transform coordinates and vectors for the moving reference frame method (!1830)

Miscellaneous

• Use std::stod instead of boost::lexical_cast (!1819)

Documentation

• Add initial documentation for the IncNavierStokesSolver (!1822)

v5.6.0

Library

• Clean-up Set_Rhs_Magnitude function in NekLinSysIter (!1729)
• Consistently use template parameters in VmathArray (!1748)
• Fix issue with CMake and zlib versions >= 1.3.0 (!1744)
• Add 1D demo and test of h-type convergence for a CG projection. (!1738)
• Add 2D projection demo and tests following 1D added in MR !1738. (!1762)
• Tidy up tolerance in NekLinSystIter and NekNonlinSysIter solvers (!1722)
• Enable varcoeffs for Collections (!1701)
• Fix misplaced " in Nektar++Config.cmake (!1742)
• Further tidy-up in linear solver (!1761)
• Use FwdTrans in UnsteadySystem when using Parareal (!1785)
• Automate deployment of README.md to dockerhub (!1786)
• Fix memory leak with Block preconditioner for time-updated matrices (!1737)
• Support for implicit sliding meshes (!1787)
• Fix compilation issue with OpenCASCADE 7.8.0 (!1799)
• Fix MPI communicator for Parallel-in-Time (!1801)
• Fix warnings issued by MSVC compiler in LibUtilities and StdRegions (!1740)
• Fix Parareal convergence monitoring (!1802)
• Avoid copying input/output to/from padded aligned arrays in MatrixFree operators(!1783)

CompressibleFlowSolver

• Complete second Frechet derivative implementation (!1761)
• Add conditional updating of elemental Mass and Laplacian matrices for LinearADR matrices (!1766)
• Added routine to order expansion in an optimal manner for MatrixFree/Collection ops (!1770)
• Fix PFASST I/O and pre-initialize coarse preconditioner for Parareal (!1749)
• Remove collection offset arrays since no longer required (!1771)
• Fix summary output (!1779)
• Update Docker images to use bookworm (!1775)
• Update clang-tidy and clang-format to v16 (!1777)
• Add coverage metric capturing (!1776)

ShallowWaterSolver

• Refractoring to reduce code duplication (!1782)

NekPy

• Add binding to NekPy to check of geometry elements are valid (!1755)
• Update NekPy to more modern packaging (!1747)
• Add wrapper for selected SolverUtils classes, particularly Filter (!1379)
• Add VTK support for OutputVtk module (!1379)
• Add bindings for SolverUtils::EquationSystem and UnsteadySystem (!1752)
• Return native Python types when getting session variables and parameters (!1380)

NekMesh

• Fix optiKind flags in VarOpti for freenodes that are on more than a single curve / surface (!1597)
• Fix VarOpti Surface Node Sliding on the CAD in 2D (!1569)
• Add feature for r-adaption on user-defined CAD curves (!1349)
• Add feature for r-adaption on user-defined CAD curves (!1349)
• Add unit testing infrustructure and initial example (!1753)
• Added a custom cmake cache file to load defaults for building only NekMesh without the solvers (!1641)
• Extend peralign module to all types of meshes (!1702)

IncNavierStokesSolver

• Matrix-Free LinearADR operator for VCSImplicit and others (!1627)
• Make substepping normal velocity evaluation more efficient (!1795)

FieldConvert

• Add range function as an option in Xml Input, align python usage and start depracation of -r option (!1791)

v5.5.0

Library

• Fix Nektar++Config.cmake to use MPI_CXX (!1224)
• Redesign of Parareal and PFASST driver (!1613)
• Update default global system solver parameters for paralell-in-time (!1649)
• Add member function in MPI communicator to check if time-parallel is enable (!1647)
• Fixed FilterError for homogeneous expansions (!1640)
• Fix ForcingAbsorption for homogeneous expansions (!1650)
• Update AssemblyMap to reduce verbosity when using parallel-in-time (!1651)
• Tidy-up of Collection library (!1622)
• Tidy-up of I/O in BasicUtils (!1623)
• Fix local implementation of CG with Null preconditioner (!1658)
• Update to use C++17 nested namespaces (!1556, !1670)
• Fix a minor bug in ProcessWallNormalData (!1663)
• Fix Explist::v_GetNormals and GetElmtNormalLength (!1625)
• Replace boost::random with std::random (!1673)
• Add a new feature of Lagrangian points tracking in parallel (!1666)
• Add Robin BC to xxt full solver (!1679)
• Update Parareal file output and tidy-up Parareal/PFASST interpolation (!1678)
• Move from boost::filesystem to std::filesystem (!1674)
• Use [[deprecated]] attribute (!1682)
• Fix QuadExp::v_ComputeTraceNormals(!1685)
• Replace boost::thread with std::thread (!1687)
• Replace boost::math::tgamma by std::tgamma(!1686)
• Remove #include <boost/math/special_functions/fpclassify.hpp> header (!1686)
• Replace boost::regex with std::regex (!1676)
• Update SDC scheme for implicit PFASST (!1659)
• Pre-allocate memory for GMRES (!1668)
• Update hdf5 read to only read in selected elements to reduce memory overhead (!1351)
• Remove arbitrary factor in GlobalLinSysIterative.cpp (!1694)
• Some tidy-up in LinearAlgebra (!1699)
• Some further tidy-up in LinearAlgebra (!1700)
• Updated hdf5 to 1.12.3 (!1696)
• Remove redundant tolerance limiter in GMRES (!1707)
• Remove unused tolerance parameter in NekSys class and subclasses (1708)
• Avoid repeatly operator assignment in NekNonlinSysNewton class (!1709)
• Add an exact solution for GetLocCoords of straight-edge quad elements (!1704)
• Add sliding mesh capability (!1605)
• Remove MaxIterations parameter from AssemblyMap (!1710)
• Consistently use relative tolerance for GMRES (!1706)
• Fix use of absolute tolerance for iterative solvers (!1711)
• Move solver info loading form NekSys (and its derived classes) to the callers (!1716)
• Tidy physderiv and helmholtz in MatrixFreeOps (!1653)
• Remove time estimation for Parareal and PFASST drivers (!1691)
• Tidy and clarify implementation of Dirichlet boundary condition in GlobalLinSys (!1724)
• Remove unused file NekLinAlgAlgorithms.hpp (!1728)

ADRSolver

• Add support for spatially-constant, but variable direction, diffusion to ADRSolver. (!1669)
• Inline Vmath library (!1667)
• Refactor UnsteadyReactionDiffusion as a subclass of UnsteadyDiffusion (!1713)
• Tidy-up ADRSolver (!1715)
• Refactor UnsteadyAdvectionDiffusion as a subclass of UnsteadyAdvection (!1720)
• Update UnsteadyReactionDiffusion solver for explicit time-stepping (!1731)
• Refactor UnsteadyViscousBurgers as a subclass of UnsteadyInviscidBurgers solver (!1718)

CardiacEPSolver

• Fix cell model history point filter output after base class change (!1342)
• Add const qualifier to SetUniversalUniqueMap (!1644)
• Add safety check for FinTime parameter for parallel-in-time (!1652)

IncNavierStokesSolver

• Save BndElmtExpansion and avoid re-building (!1648)
• Add Simo-advection and a switch for Simo-/Dong-advection to VCSImplicit (!1630)
• Add a new feature of elasticaly mounted object using the moving reference frame (!1495)
• Some tidy-up (!1693)
• Fix memory leak in VCSImplicit due to matrix updating (!1688)
• Rename Simo-/Dong-advection to Extrapolated/Updated respectively for VCSImplicit (!1717)
• Update test for SuccessiveRHS parameter (!1727)
• Fix segmentation error in IncNavierStokesSolver when AeroForces filter is used (!1726)
• Added Examples in solvers/IncNavierStokesSolver/Examples, matching with the user-guide (!1723)
• Extend support for IO timer levels to IncNavierStokesSolver (!1732)
• Fixed issue with substepping when using mixed BCs (!1639)