AlexanderMueller authored 3 years ago and Sander, Oliver committed 3 years ago

This commit adds the energy implementat for the nonlinear Reissner-Mindlin
shell model known as the Simo and Fox shell model.

Details may be found in

  "A consistent finite element formulation of the geometrically
   non-linear Reissner-Mindlin shell model [Müller, Bischoff]"

With this, the preparatory stretching experiment for film-on-substrate can be done using dune-gfe.
The vertices for adaptive refinement can now be selected without specifying a shell area.

Correct the sign of the Neumann and volume forces to make it consistent throughout dune-gfe and dune-elasticity

In the energy file, the energies are *added*.
The Neumann and volume forces get the correct sign in the actual program file by mulitplying with (-homotopyParameter).

WHITESAPCES for previous commit: Add the option to use the Riemannian Proximal Newton solver also for problems with dim = dimwold

Use cholmod if the dune-solvers version >= 2.8, use umfpack for older versions

Change the keywords for switching between the solvers to trustRegion and proximalNewton also in film-on-substrate

Now all combinations of dim = dimworld or dim != dimworld and MIXED_SPACE = 0 or MIXED_SPACE = 1 at least compile.

This has to be dimworld and not dim.

The StressFreeStateGridFunction is the actual parametrization from the grid to the
nonplanar Cosserat shell in stress-free state.
The geometries of the StressFreeStateGridFunction will then be used for all calculations
instead of the linear geometries of the piecewise linear grid.

This has been changed to Dune::Vtk::FormatTypes::ASCII in the dune-vtk module

The previous implementation worked only for first-order finite element
spaces, because it assumed that grid vertices and Lagrange nodes
were identical.

So far, the Cosserat rod energy implementation hat a first-order
finite element space hardcoded.  This patch removes that restriction.
As for the other models in this Dune module, the finite element basis
is now a template parameter of the model energy, and can be set to
any reasonable basis.

It is not a 'stiffness' anymore, but really only implements the
energy.

This is a prerequisite to a) support for higher-order FE,
and b) getting rid of the hand-written CosseratVTKWriter.

Previously, the rod3d code used a hand-implemented first derivative
of the rod energy, and FD for the second derivatives.  This patch
replaces this by ADOL-C for both first and second derivatives.
Advantages are:
* The code is much shorter, and easier to understand.
* The previous code contained interpolation formulae for
  first-order 1d geodesic finite elements.  This restricted
  rod problems to first-order GFE.  A follow-up patch will
  allow higher-order approximations.
* The new code runs roughly twice as fast.

GeodesicFEAssembler can replace it without problems.

In the 'mixed' case the volume term was simply forgotten, and not
actually handed to the assembler.

It is very unlikely that somebody will use that ever again.

Rather than using a default file, which was the case previously.

Otherwise the program aborts for the usual mysterious reasons.

When dune-curvedgeometry is not found, the normal Derivative is set to zero.