MFront's computations are absolute, and incorporation
of initial stresses requires computation of initial
strain.

Fix docu

Internal variables for TM/RM/HM

Repeat time step on assembly error

Needs output file name format which does not include
the time step number. Then a comparison on different
test environments will be possible.
Currently it fails because some platforms need say 400
time steps but other 398.

This causes the time step to fail and repeated if possible.

Original test written by Vinay Kumar.

3D mesh to grid conversion

Biot effect LIE/HM

[web] Updated Hugo to 0.60.1

Adapt Robin / Neumann boundary conditions to user needs.

Relax the test tolerances for the parallel TracerSimulation benchmark.

For the first timestep the initialization would fail
if the property is defined on nodes.
It was not caught before because the LIE/HM/single_fracture_3compartments_flow_linear_aperture0.prj
test didn't use the right material_id value for the
fracture.

The Darcy velocity wasn't initialized at all and lead
to spurious numbers in output when the flow in the
matrix was deactivated.

Thomas Nagel authored 5 years ago and Dmitri Naumov committed 5 years ago

Added roller bc.

Fixed a bug in MeshComponentMap under PETSc with single thread