The adaptivity is borken, see FS#237.
(reviewed by bernd)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13344 2fb0f335-1f38-0410-981e-8018bf24f1b0

Make complained about an empty variable. Probably it did
not like the :=+, thus it was replaced by " \" and
everything is added in one line.
(reviewed by fetzer)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13341 2fb0f335-1f38-0410-981e-8018bf24f1b0

Some more places are affected that where not included in last commit.
(reviewed by fetzer)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13339 2fb0f335-1f38-0410-981e-8018bf24f1b0

(reviewed by natalies)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13328 2fb0f335-1f38-0410-981e-8018bf24f1b0

It was never officially part of the interface and was
removed after Dune 2.3.
Replaced by global.
(reviewed by martinb)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13323 2fb0f335-1f38-0410-981e-8018bf24f1b0

Use Entitiy.level() which is in most cases simply
EntityPointer->level().
(review by fetzer)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13322 2fb0f335-1f38-0410-981e-8018bf24f1b0

I don't know why it is needed now and was not yet needed.
(reviewed by fetzer)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13320 2fb0f335-1f38-0410-981e-8018bf24f1b0

Thanks to Georg Futter for reporting this.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13313 2fb0f335-1f38-0410-981e-8018bf24f1b0

The two-phase geomechanics model is based on PDELab 2.0 now. In order to
avoid patching, the ElTwoPModel is based on the new ElTwoPBaseModel
instead of the ImplicitModel. The ElTwoPBaseModel is almost a copy of
ImplicitModel and only differs in the way that the SolutionVectors are
handled. There's definitely room for improvement here. 



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13296 2fb0f335-1f38-0410-981e-8018bf24f1b0

Since everything has been moved to PDELab 2.0, the checks for a patched
PDELab are no longer needed. 



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13295 2fb0f335-1f38-0410-981e-8018bf24f1b0

If the CO2NI model is used, the the properties IsothermalModel and
IsothermalVolumeVariables have to be set in the corresponding problem
file.



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13294 2fb0f335-1f38-0410-981e-8018bf24f1b0

There appears to be an incompatibility between clang and gcc <= 4.6
regarding friend declaration combined with typedef. This should be a
temporary workaround until a proper test is written (or found in Dune).



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13293 2fb0f335-1f38-0410-981e-8018bf24f1b0

git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13292 2fb0f335-1f38-0410-981e-8018bf24f1b0

models to the generic setting

In order to use a new nonisothermal model, only the includes in
the corresponding application files have to be changed from
including the dedicated non-isothermal model such as
#include <dumux/implicit/2p2cni/2p2cnimodel.hh>
to the corresponding isothermal one such as
#include <dumux/implicit/2p2c/2p2cmodel.hh>
This has been done for the tests in test/implicit/2pni, ...2p2cni and
...co2ni.

For enabling the isothermal models to get the new non-isothermal
functionality, it was sufficient to modify the files ...properties.hh
and ...propertydefaults.hh:

- In ...properties.hh, a non-isothermal type tag such as "TwoPTwoCNI" is
  introduced that inhereits from the corresponding isothermal one AND
  the type tag "NonIsothermal". In order to not mix it up with the old
  non-isothermal models, corresponding safety guards were added.

- In ...propertydefaults.hh, the properties introduced in
  niproperties.hh of the isothermal model have to be set.

The old non-isothermal models have been deprecated by emitting a warning
if the corresponding ...properties.hh file is included. To ensure that
they still work, the ...properties.hh file is included on top of every
file that corresponds to the old model. Together with the safety guard
mentioned above, this guarantees that in fact the old model is chosen if
desired.

Reviewed by Alex K.



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13290 2fb0f335-1f38-0410-981e-8018bf24f1b0

conductivities, correct comments and spacing.

The additional properties required by the generic non-isothermal model
can be centralized in a file niproperties.hh. 

The signature of the effectiveThermalConductivity functions is unified.
They get the volume variables, spatial parameters and the arguments that
will be required for calling the spatial parameters. When attempting to
use the nonisothermal model for 3p3c, it will be seen whether this
signature is genral enough.

Reviewed by Alex K.



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13289 2fb0f335-1f38-0410-981e-8018bf24f1b0

files and installation script



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13288 2fb0f335-1f38-0410-981e-8018bf24f1b0

Allows to use an unpatched (!) PDELab 2.0 for the multidomain
applications. Users have to upgrade to PDELab 2.0 and the 2.0 release
branch of dune-multidomain. 

- Subdomain GridFunctionSpaces are expected to use
  Dune::PDELab::NoConstraints instead of
  ...NonoverlappingConformingDirichletConstraints. Users have to remove
  the corresponding property settings in the subproblem files.

- Subdomain GridOperators are no longer needed. The property name has
  been removed from subdomainproperties.hh. Users have to remove
  the corresponding property settings in the subproblem files.

- Use OrderingTags instead of Mappers.

- The storage of residual, current and old solution in the
  assembler/model uses shared_ptr now. 

- The components of the multidomain vector containers are accessed via
  the base() method. Everything in splitandmerge.hh still relies on
  EntityBlockwiseOrdering. 

Reviewed by gruenich.



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13287 2fb0f335-1f38-0410-981e-8018bf24f1b0

Reviewed by gruenich.



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13286 2fb0f335-1f38-0410-981e-8018bf24f1b0

Reviewed by gruenich.



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13285 2fb0f335-1f38-0410-981e-8018bf24f1b0

This merges the improvements to the AMG backend from the strip-pdelab
branch. Improvements are twofold:

1. The AMG backend does not require PDELab anymore. It is based directly
on dune-istl.

2. The convergence behavior is improved for parallel runs, at least for
the cases tested so far.

Thanks to Markus Blatt for the work.



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13284 2fb0f335-1f38-0410-981e-8018bf24f1b0

This improves the behavior of GET_PARAM.

Thanks to Markus Blatt for the work.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13283 2fb0f335-1f38-0410-981e-8018bf24f1b0

correct use of wildcards, add ALUGrid flags to tests with AMG.

Merge revisions 12949, 13246 and 13248 from the strip-pdelab branch.
They all improve the buildsystem behavior.

Thanks to Markus Blatt for the work.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13282 2fb0f335-1f38-0410-981e-8018bf24f1b0

git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13281 2fb0f335-1f38-0410-981e-8018bf24f1b0

Note: It is not clear to me we use atOnceAccu and not
successiveAccu. Anyway as long as we use less than 100 processes
this should not bite us too much.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13269 2fb0f335-1f38-0410-981e-8018bf24f1b0

Previously this was done in every apply call. Unfortunately, the
initial condition for the algorithm was only satisfied in the first call.
This patch make sure that initGhostsAndOwners is only called during the
first apply method.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13266 2fb0f335-1f38-0410-981e-8018bf24f1b0

- The implicit 1p2c model is ready to use with the non-isothermal model
- Added 1p fluidmatrixinteractions to calculate effective thermal conductivity of fluid and solid
- Adjusted test/implicit/1p2c:
	- 1p2coutflowproblem.hh can be switched to nonisothermal
	- 1p2cconductionproblem.hh compares the model results to a simple analytical model
	- 1p2cconvectionproblem.hh compares the model results to another simple analytical model

Reviewed by Bernd

git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13264 2fb0f335-1f38-0410-981e-8018bf24f1b0

The change to conservation of moles in the continuity balance brought
some non-fuzzy, but tolerable changes, to the Stokes2c(ni) tests.
Moreover, the stabilization parameter for the Stokes2c test has been
increased to dampen further the unphysical pressure oscillations.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13263 2fb0f335-1f38-0410-981e-8018bf24f1b0

Undefined values have been accessed in the StokesncniLocalResidual.

Reviewed by Thomas.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13261 2fb0f335-1f38-0410-981e-8018bf24f1b0

Stokes2c(ni) has been deprecated in 2.5. After some issues according to
FS#235 have been clarified, it can be removed.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13259 2fb0f335-1f38-0410-981e-8018bf24f1b0

For some situations, it appears to be useful to use the molar instead of
the mass density in the Stokes continuity equation. In order to enable
this behavior, the property UseMoles is used for Stokes and the
VolumeVariables are equipped with the required functionality.
Corresponding parts can be deleted from the derived Stokesnc model.

Modified from a patch provided by Vishal, thanks.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13258 2fb0f335-1f38-0410-981e-8018bf24f1b0

of FluidSystem

This is consistent with the standard local residuals. 

Patch provided by Vishal, thanks.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13257 2fb0f335-1f38-0410-981e-8018bf24f1b0

git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13253 2fb0f335-1f38-0410-981e-8018bf24f1b0

This patch fixes various bugs in the non-pdelab implementation:
1. The index sets are now correctly setup.
2. For the box model it is made sure that the right hand side is
consistent.

Now the test_box1pwithamg converges in the same number of steps of
the non-linear solve in parallel as in serial. (Not the case with
pdelab!)


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13252 2fb0f335-1f38-0410-981e-8018bf24f1b0

Patch provided by Bernd Flemisch.

Implements the datahandles completely and correctly and adds
missing call to initGhostsAndOwners for the nonoverlapping case.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13251 2fb0f335-1f38-0410-981e-8018bf24f1b0

We use the normal initializers of the class constructor for this now.

NOTE: The member will be uninitialized if the default constructor is called!


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13250 2fb0f335-1f38-0410-981e-8018bf24f1b0

The compiler error was:
/home/mblatt/src/dune/release/dumux/dumux/decoupled/2p/diffusion/fvmpfa/lmethod/fvmpfal3dinteractionvolumecontaineradaptive.hh:96:5: error: a class-key must be used when declaring a friend
/home/mblatt/src/dune/release/dumux/dumux/decoupled/2p/diffusion/fvmpfa/lmethod/fvmpfal3dinteractionvolumecontaineradaptive.hh:96:5: error: friend declaration does not name a class or function
make[4]: *** [test_adaptive2p2c2d.o] Fehler 1

Seems like g++-4.6 is having problems with typedefs in a friend
declaration. We now us the original class type for this.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13249 2fb0f335-1f38-0410-981e-8018bf24f1b0

The correct usage of wildcards is "$(wildcard *.hh)".
With this patch we use these instead of "*.hh".

Quoting from https://www.gnu.org/software/make/manual/html_node/Wildcard-Examples.html#Wildcard-Examples

Wildcard expansion does not happen when you define a variable. Thus, if you write this:

     objects = *.o

then the value of the variable objects is the actual string ‘*.o’. However, if you use the value of objects in a target or prerequisite, wildcard expansion will take place there. If you use the value of objects in a recipe, the shell may perform wildcard expansion when the recipe runs. To set objects to the expansion, instead use:

     objects := $(wildcard *.o)



git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13248 2fb0f335-1f38-0410-981e-8018bf24f1b0

We need to include $(top_srcdir) in AM_CPPFLAGS for this to work.


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/branches/strip-pdelab@13246 2fb0f335-1f38-0410-981e-8018bf24f1b0

git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13242 2fb0f335-1f38-0410-981e-8018bf24f1b0

writer: call that function in order to write it to result file
reference: identical except one additional variable present


git-svn-id: svn://svn.iws.uni-stuttgart.de/DUMUX/dumux/trunk@13234 2fb0f335-1f38-0410-981e-8018bf24f1b0