Several material properties, corresponding the the tetrahedra material flag, can now be defined in the data file. For example :

[./physics]
material_flag   = 1,2               # 2 different materials, tetrahedra flags 1 ou 2
young           = 3.0E6,1.9E9       # 1 -> 3.0e6, 2 -> 1.9e9
poisson         = 0.3,0.3	    # 1 -> 0.3,   2 -> 0.3

The previous declaration is still available :

[./physics]
material_flag   = 1
young           = 3.0E6
poisson         = 0.3

if we want to use only one material.

added new () operator that returns a boolean argument describing if the returned value is the default value or not in order to make MLPreconditioner working again

A small change due to the new GetPot syntax (as described previously in the mailing list we have to switch from char* to string).

Moreover some new functions have been added: in particular I have added a new function which returns true if a variable has been defined in the GetPot data file; otherwise it returns false.

Look inside the GetPot.hpp file to have more information about that.

the HarmonicExtensionSolver::updateDispDiff() method was previously included into HarmonicExtensionSolver::iterate()

restyling of VenantKirchhof: the methods that are specific for the monolithic solver are moved to the class Monolithic. The variables that are not needed in the monolithic case are not initialized pointers.

moving the preconditioner for Oseen into the linear solver (SolverTrilinos), commented the construction of precLin, preconditioner for the Jacobian system, that is not used

restyling HarmonicExtensionSolver: added the method applyBoundaryConditions(..), added the possibility to set an offset, as. The method updateDispDiff() was previously included in iterate()

moving the preconditioner to the class SolverTrilinos, the matrix on which we want to build the preconditioner is passed to solveSystem

Resolved with Gilles

Now corrected

In solveSystem, the linear system is set up with a given preconditioner,
it is solved, and (if retry = true) in case of non-convergence, the preconditioner is
recomputed. (cf doc in SolverTrilinos.hpp)

Also:
- changed cout in Display.leaderPrint.
- In Oseen the first resolution was done without preconditioner. Corrected.

changed width to  << setw(int).
removed close() at the end of the methods (the file is closed anyway, but I wonder why it is
not working with close() )

due to the change in PostProc method average, it is now mandatory to specify that average method will work on the finite element space "1" (being 0 - velocity feSpace, 1 - pressure feSpace)

moved default parameters from function definitions to function declarations (compatibility with intel c++ compiler)

+ the default "feSpace" parameter for "average" method is changed (0 instead of 1).

I have checked, but I will recheck after Paoloś commits