Changes between version 1.0 and version 1.1

This is the list of changes made after the release of Aspect version 1.0 for version 1.1. All entries are signed with the names of the author.

  1. Changed: the Stokes linear solver tolerance now includes the current pressure instead of being only based on the rhs. This makes the tolerance less problem dependent. Please note that in many cases you can loosen the tolerance to something between 1e-4 to 1e-7.
    (Timo Heister, 2014/05/30)

  2. New: ASPECT now logs all screen output to output/log.txt.
    (Timo Heister, 2014/05/29)

  3. Changed: The adiabatic reference profile is now defined in a plugin architecture as well. Currently, only the old funtionality of having a constant profile, which is computed before the first timestep, is included. However, now there is the possibility for more simple or more realistic user-written plugins (like an analytic function, or a profile that is updated over time).
    (Rene Gassmoeller, 2014/05/29)

  4. Changed: The initial conditions do not longer use pointers to other parts of the program (like the geometry model), instead they now should be derived from the SimulatorAccess class as all other plugins.
    (Rene Gassmoeller, 2014/05/29)

  5. Changed: For some modules, e.g. the selection of which postprocessor to run, "all" was an option. This may have made sense when we had 3 or 4 such postprocessors, but this is no longer the case. Furthermore, not all postprocessors can always be run for a given geometry, input parameters, etc. Consequently, the option of specifying "all" has been removed. Where this applied, we have also removed the default value "all".

  6. Changed: If the input file is invalid, for example because a parameter's value did not satisfy its constraints or because it tried to define a parameter that did not exist, all processors in a parallel computation produced the same error message – leading to massive amounts of output that were barely readable. This has been changed now: only processor zero now generates the output, but all processors abort nonetheless.
    (Wolfgang Bangerth, 2014/05/29)

  7. Changed: Previously, we aborted the program if the specified output directory did not exist. This is inconvenient for jobs that may have been sitting in a queue for a long time and for which forgetting to create the output directory may have just been an oversight. We now simply generate this directory if it is missing.
    (Wolfgang Bangerth, 2014/05/28)

  8. New: Given the proliferation of plugins of all kinds, plugin names are now listed in alphabetical order in the documentation.
    (Wolfgang Bangerth, 2014/05/27)

  9. New: PETSc support can now be enabled using 'cmake -D ASPECT_USE_PETSC=ON'.
    (Timo Heister, 2014/05/27)

  10. Fixed: The GPlates plugin now correctly handles meshes created by GPlates 1.4 and later. Previous Aspect versions may only read in files created by GPlates 1.3.
    (Rene Gassmoeller, 2014/05/23)

  11. Add a find_postprocessor() function in Simulator_access class to pass a pointer of a certain postprocessor, if it is not found, return a NULL pointer.
    (Siqi Zhang, 2014/05/23)

  12. Mesh refinement plugins can now interact with initial global mesh refinement and unflag certain cells if desired.
    (Timo Heister, 2014/05/22)

  13. Changed: The functionality that outputs some basic statistics values like the Rayleigh number in case of a simple material model was moved from the velocity statistics postprocessor to a new postprocessor called basic statistics.
    (Rene Gassmoeller, 2014/05/21)

  14. Fixed: We accidentally evaluated the viscosity of the material model from the place where we compute the adiabatic conditions, but this required information that we didn't have. This is also unnecessary since we don't need the viscosity in this place. This is now fixed.
    (Wolfgang Bangerth, 2014/05/21)

  15. Fixed: Temperature and compositional boundary conditions were previously evaluated only once at the beginning of the simulation, but this did not allow for time dependent boundary conditions for these variables. This is now fixed.
    (Wolfgang Bangerth, 2014/05/21)

  16. Fixed: We forgot to set the initial time before we evaluate the temperature boundary conditions, so temperature boundary conditions could not use get_time() – they just got NaN. This is now fixed.
    (Wolfgang Bangerth, 2014/05/21)

  17. New: There is now functionality for running models with a free surface with an Arbitrary Lagrangian-Eulerian framework. The user specifies which boundary indicators are to be free surface boundaries, as well as a stabilization parameter that keeps the free surface from exhibiting a sloshing instability. There is an associated postprocessor that calculates mininum/maximum topography, as well as a new section and cookbook in the manual. Finally, there are two tests for the free surface formulation.
    (Ian Rose, 2014/05/21)

  18. New: There is now a new postprocessor "velocity boundary statistics" that computes statistics about the minimal and maximal velocities on each part of the boundary.
    (Anne Glerum, 2014/05/21)

  19. New: Compositional fields have names now. If no names are specified in the input file, the previously used names C_1, C_2, ... are used as a default.
    (Juliane Dannberg, 2014/05/21)

  20. New: There is now a simple compressible material model with constant compressibility resulting in an exponential dependeny of density on pressure and a linear dependence on temperature deviation from the adiabatic profile. All other material properties are constant.
    (Rene Gassmoeller, 2014/05/20)

  21. New: There is now a new initial temperature condition plugin that can read in solidus temperatures from a file and add perturbations to it. There is also a corresponding data file for initial conditions for Mars from Permentier et al. in data/initial-temperature/solidus.Mars.
    (Siqi Zhang, 2014/05/20)

  22. New: There is now a new postprocessor "spherical velocity statistics" that computes statistics about the radial and tangential velocity field.
    (Anne Glerum, 2014/05/20)

  23. New: There is now a mesh refinement criterium that checks the mesh always has a minimum refinement level determined by a function given in the input file.
    (Juliane Dannberg, 2014/05/20)

  24. New: Input files can now contain lines ending in backslashes to concatenate subsequent lines.
    (Menno Fraters, 2014/05/19)

  25. New: There is a new material model called multicomponent for having an arbitrary number of compositional fields with different material properties, where each compositional field represents a rock type. Within each rock type the material properties are assumed to be constant. When more than one rock type is present, the material model averages their properties with a weighted arithmetic average. An exception is viscosity, where the user may specify a harmonic, geometric, or arithmetic averaging, or selecting the viscosity of the maximum composition.
    (Ian Rose, 2014/05/19)

  26. New: The dynamic topography postprocessors now take into account (i) the dynamic pressure, (ii) effects due to compressibility, (iii) they now have the option to subtract the mean topography to ensure that the topography is, on average, zero.
    (Jacqueline Austermann, 2014/05/19)

  27. Fixed: The GPlates cookbook in Aspect-1.0 contained a bug that prevented it from using the GPlates plugin. This is fixed now.
    (Rene Gassmoeller, 2014/05/18)

  28. New: There is a new plugin architecture for heating models that allows for plugins defining the radiogenic heating rate in dependency of temperature, pressure, composition, position and time. This introduces an incompatibility of input files to older versions, but the old functionality is preserved and extended.
    (Rene Gassmoeller, 2014/05/16)

  29. New: There is now a material model with reactions between compositional fields and a cookbook describing the usage of this material model.
    (Juliane Dannberg, Rene Gassmoeller, 2014/05/16)

  30. Fixed: Times associated with visualization output are now correctly modified when "Use years in output instead of seconds" is true.
    (Eric Heien, 2014/05/09)

generated on Tue Jan 24 2017 19:53:10 by doxygen
Hosting by IWR Universit├Ąt Heidelberg