Minor API extension

API additions

• Added "AllocateElemmat".

Other updates

• Updating docs.

Small changes all around

API additions

• Added "AllocateQtensor", "AllocateQscalar", "AllocateDofval", "AllocateNodevec", "AllocateElemvec"

API changes

• Removing Paraview interface: replaced by external libraries "XDMFWrite_HighFive" and "XDMFWrite_h5py"

• Element*: "dV" now only returns raw data, the "asTensor" member function (and free function) can be used to convert the 'qscalar' to a 'qtensor'

• Separating sparse solver in separate class to offer more flexibility in the future

• Adding "dot" to "Matrix"

Other updates

• Applying clang-format to source, python API, tests, and examples.

• Adding test GMatElastoPlasticQPot

• Adding test based on hybrid material definitions

• Formatting update: renaming all return variables "out" to "ret"

• Correction zero allocation to allows for dofval.size() > nodevec.size()

• Formatting update xt::amax and xt::sum

• Renaming private function to begin with caps when the function allocates its return data

• Reducing copies when using Eigen

• Reducing default size examples

• Supporting Windows (#87)

• Removing xtensor_fixed

• Using xt::has_shape

New function

• Adding function to compute the edgesize of all elements

Modern CMake

• Switching to modern CMake
• Introducing extra build targets
• Starting integrating examples in CI

Full documented version

• Added minimal documentation.
• Completed Python interface.
• Added ParaView/HDF5 support.
• Added Eigen's solver concept.

Code-style overhaul; added tyings

API changes:

• All member functions that wrap other member functions to allocate data now start with a capital.
• Matrix....solve(...) no longer modifies the right-hand-side. Use Matrix....reaction(...) function to get the right-hand-side that belong to prescribed DOFs (i.e. reaction forces).
• quad.dVtensor is deprecated over quad.DV(rank).

Extensions:

• The headers TyingsPeriodic, VectorPartitionedTyings, and MatrixPartitionedTyings now provide functionality to run periodic simulations.

Part of the code now parallel

The code for "Dynamics" has been completely restructured. The implementation now features:

• Immediate decoupling of elements, storing intermediate variables such as the shape function gradients (allowing later re-use).
• Almost everything can now be run in parallel.

First working version

First working version, transferred from old modules.