Update of the shape derivative terms for Newton FSI. Now for P1P1 stabilized elements the Jacobian matrix is explicitly computed (not just as matrix-vector multiplication). This allows to compute the shape derivatives once per Newton iteration (instead of once per Jacobian times vector multiplication).

update of the shape derivative terms for Newton FSI. Now for P1P1 stabilized elements the Jacobian matrix is explicitly computed (not just as matrix-vector multiplication). This allows to compute the shape derivatives once per Newton iteration (instead of once per Jacobian times vector multiplication).

cleaned datafile. Still a lot of cleaning needed.

are read by default by SolverTrilinos

using correct method to see if the preconditioner is set

type of importer/exporter in data file

Removing prefix inside HDF5 files. It is not necessary, and prevents common color scale when dealing with more that one model in the same output.

for P1Bubble the problems was in the xmf description file, not in the h5

Removed a warning + added output information + added default ./ folder for output files (it was just for testing it)

By default, stiff strain is turned off. To enable it put in the data file the following line:

stiff_strain    = true

in the [space_discretization] section of the GetPot file.

It is also possible to specify (as for ensight) a directory for output files.

NOTE: seems that the exporter is not running with P1Bubble-P1 and P2-P1 FE. It works only with P1-P1.

bug solved when saving only every n>1 timesteps

reuse prec and max iter for reuse have to be set in the datafile:
    M_maxIterForReuse = dfile( ( section + "/max_iter_reuse").data(), M_maxIter*8/10);
    M_reusePreconditioner = dfile( (section + "/reuse").data(), M_reusePreconditioner);

reuse stabilization is in the data file, ipstab section
    max_iter_reuse  = 60

In particular removed  solveSystem which i now in SolverTrilinos.
Moreover the data file should be modified accourding to:

reuse prec and max iter for reuse have to be set in the datafile:
    M_maxIterForReuse = dfile( ( section + "/max_iter_reuse").data(), M_maxIter*8/10);
    M_reusePreconditioner = dfile( (section + "/reuse").data(), M_reusePreconditioner);

In particular removed  solveSystem which i now in SolverTrilinos.
Moreover the data file should be modified accourding to:

reuse prec and max iter for reuse have to be set in the datafile:
    M_maxIterForReuse = dfile( ( section + "/max_iter_reuse").data(), M_maxIter*8/10);
    M_reusePreconditioner = dfile( (section + "/reuse").data(), M_reusePreconditioner);

This class was not migrated before, not it should be all right

In particular removed  solveSystem which i now in SolverTrilinos.
Moreover the data file should be modified accourding to:

reuse prec and max iter for reuse have to be set in the datafile:
    M_maxIterForReuse = dfile( ( section + "/max_iter_reuse").data(), M_maxIter*8/10);
    M_reusePreconditioner = dfile( (section + "/reuse").data(), M_reusePreconditioner);

Cleaned name of methods and removed repeated methods

removed template for solveSystem and computePrec

solveSystem has two versions:
- called with a matrix to compute the preconditioner
- called with an already computed preconditioner

reuse prec and max iter for reuse have to be set in the datafile:
    M_maxIterForReuse = dfile( ( section + "/max_iter_reuse").data(), M_maxIter*8/10);
    M_reusePreconditioner = dfile( (section + "/reuse").data(), M_reusePreconditioner);

solveSystem and solve take the rhs as a const, but in solve there is a const_cast since
the Solver from Aztecoo takes a non const (because of rescaling?)
We should be careful if you use scaling

Re-enabling a stabilitazion term. By default it is not used. So no changes if you don't turn on it from the file.

exporter has an extra method to know if the exporter want a repeated or unique map

added extra method to now if the exporter want a repeated or unique map

hdf5 now can import

test_filters has been updated and will soon be added to the testsuite

in principle, should be add, but for importing is useful to have Zero

 - Removed old unused functions
 - Split AztecOO solver from AztecOO (unused) preconditioner parameters.

AztecOO preconditioner will be placed in a AztecOOPreconditioner class in the future (now is under testing)