Changes after the latest release (v2.0.0)

This is the list of changes made after the release of Aspect version 2.0.0. All entries are signed with the names of the author.

1. Changed: If the 'prescribed field' compositional field method is used for a field, this field will now be solved in the order it is listed in in the input file (as is done for all compositional fields) instead of being updated after all of the other fields are solved (the previous behaviour).
   (Juliane Dannberg, 2018/10/31)

2. New: There is now a new method for compositional fields that allows to prescribe them to a value that is computed in the material model and then, in a second step, to diffuse them based on a length scale given as an input parameter. This option can be used by selecting the 'prescribed field with diffusion' compositional field method in the input file, and allows it to smooth quantities that are computed in the material model.
   (Juliane Dannberg, 2018/10/29)

3. Fixed: The 'named additional outputs' postprocessor used to only support a single named output from the material model due to an indexing bug. This is fixed now.
   (Rene Gassmoeller, 2018/10/04)

4. New: Compositional fields can now be prescribed to a value that is computed in the material model as an additional output at every time step and then interpolated and copied into the solution vector (as opposed to advecting the compositional field). This option can be used by selecting the 'prescribed field' compositional field method in the input file.
   (Juliane Dannberg, 2018/10/01)

5. Improved: The artificial diffusion term that is added in the entropy viscosity method to the temperature and composition equations is now computed as the maximum of the physical diffusion and entropy viscosity instead of the sum. This is reasonable, as the entropy viscosity is the smallest diffusion that is necessary to stabilize the equation, and the presence of physical diffusion stabilizes the equation exactly as an artificial diffusion would. This change likely changes all benchmark results, but leads to more accurate, less diffusive solutions. We verified the new method in selected benchmarks, and updated test results but did not rerun and update all of the contained benchmark cases.
   (Wolfgang Bangerth, Rene Gassmoeller, 2018/09/26)

6. Fixed: In case of combining reference profile formulations of the equations (ALA, Boussinesq) with discontinuous temperature elements the face terms in the assembly were incorrectly assembled with the full, instead of the reference density. The volume terms were correct. The result were temperature overshoots and possible crashes if the face terms dominate and create densities smaller than zero. This is fixed now, the face terms are now correctly computed with the reference density.
   (Rene Gassmoeller, 2018/09/15)

7. Fixed: The global entropy variation that is used in the artificial viscosity computation was always computed with a quadrature of the polynomial degree of the temperature (even for compositional fields). This is fixed now, and causes minor changes in the stabilization for compositional fields if they use a different element order than the temperature (usually not the case).
   (Rene Gassmoeller, 2018/08/31)

8. New: There is now a 'material statistics' postprocessor that computes the average density, average viscosity, and total mass in a model.
   (Rene Gassmoeller, 2018/08/22)

9. New: ASPECT has two visualization postprocessors which calculate and output the grain lag angle and the infinite strain axis (ISA) rotation timescale, respectively. These two quantities can be used to calculate the grain orientation lag parameter of Kaminski and Ribe (G3, 2002)
   (Hannah Mark, 2018/08/12)

10. Fixed: Fixed a bug in the 'Named additional outputs' postprocessor, which would cause a segmentation fault if more than one named additional material output object was active in the same model.
    (Rene Gassmoeller, 2018/08/06)

11. Fixed: Fixed a bug in the 'Steinberger' material model that caused ASPECT to use wrong compressibilities when only a single compositional field, and a single material data file were prescribed. The correct behavior would have been to ignore the compositional field and use the data file as background material, but instead the compositional field was interpreted as volume fraction and multiplied with the compressibility calculated from the data file. This is fixed now.
    (Rene Gassmoeller, 2018/07/29)

12. New: ASPECT now outputs a dynamically generated URL based on used features to ask people to cite appropriate papers.
    (Timo Heister, 2018/07/25)

13. New: Added a postprocessor for the temperature anomaly, defined as temperature - (depth average of temperature).
    (Max Rudolph, 2018/07/11)

14. New: ASPECT now has a Patch on S40RTS initial temperature model which combines S40RTS and an ascii grid containing an upper mantle tomography. Data is input in Vs and converted to temperature within the function (using the same method as S40RTS_perturbation).
    (Sophie Coulson, 2018/07/09)

15. New: Analytical benchmark for 2D Cartesian incompressible Stokes flow following the FEM book by Donea and Huerta.
    (Cedric Thieulot, 2018/07/02)

16. New: The benchmarks now include a viscoplastic thermal convection benchmark as described and performed in Tosi et al. (G3 2015).
    (Anne Glerum, 2018/06/27)

17. Impoved: The assembly speed for the Newton solver has been significantly improved by caching various variables.
    (Menno Fraters and Rene Gassmoeller, 2018/06/26)

18. Fixed: The default latitude and longitude limits for the 'uniform radial' particle generator were set to (0, pi) even though they are measured in degrees. The new default limits correspond to the entire surface of the sphere, and the manual clarifies that latitude is in (0, 180) and not (-90, 90).
    (Bart Niday, 2018/06/25)

19. New: ASPECT can now call PerpleX to calculate material properties, phase amounts and compositions on-the-fly. A simple material model (PerpleX lookup) is provided which uses PerpleX to provide densities, compressibilities, thermal expansivities and heat capacities on a cell-by-cell basis. This model is provided as a proof-of-concept; more efficient procedures will be required for production runs.
    (Bob Myhill, 2018/06/25)

20. New: We now have unit_tests in the unit_test/ folder using the Catch testing framework. They should be used instead of writing a test in tests/ if they don't require an ASPECT run.
    (Timo Heister, 2018/06/24)

21. New: The ascii data and function boundary velocity plugins now allow velocities to be specified along spherical (up, north, east) unit vectors.
    (Bart Niday, 2018/06/24)
    New: Added a visualization plugin that directly outputs the strain rate tensor.
    (Bart Niday, 2018/06/24)

22. New: ASPECT can now read in a depth-dependent initial temperature from ASCII data. This is intended to provide a background temperature profile on which to add perturbations for composite initial temperature models.
    (Marie Kajan, 2018/06/24)

23. New: ASPECT can now read in a depth-dependent vs to density conversion file, which can be used for S40RTS and SAVANI.
    (Jacky Austermann, 2018/06/24)

24. New: Function GeometryModel::Interface::cartesian_to_other_coordinates converts between defined coordinate systems. This might break some plugins, which can be fixed by including aspect/geometry_model/interface.h
    (Bart Niday, 2018/06/23)

25. Changed: We now only set time step number and timestep sizes after initialization is finished.

    Right now, there is no way to find out, for example in the hook that is called from the function that builds constraints, whether we are already in the process of time stepping, or only in the initialization phase. So move the initialization of the time step number and time step variables to the very end of the initialization phase. Before this, these variables have invalid values.
    (Wolfgang Bangerth, 2018/06/20)

26. New: There is now a visualization postprocessor that outputs the compaction length, the characteristic length scale of melt transport.
    (Joe Schools, 2018/06/19)

27. Fixed: When gnuplot was used as the output format for the 'depth average' postprocessor plugin, the first two columns were labeled 'x' 'y', even though 'depth' and 'time' are always the first two variables. This is now fixed provided deal.II 9.0 or newer is used.
    (Bob Myhill, 2018/06/19)

28. New: Compositional fields can now be advected with the melt velocity (as opposed to the solid velocity), and this option can be used by selecting the 'melt field' advection method in the input file.
    (Juliane Dannberg, 2018/06/19)

29. New: ASPECT now does some basic checks that essential parameters (such as the number of compositional fields) are the same before creating a checkpoint and after restarting from it.
    (Wolfgang Bangerth, 2018/06/18)

30. Fixed: The 'compositional heating' heating plugin had a parameter 'Use compositional field for heat production averaging' that was used inconsistently with its description. Its first entry did not correspond to the background field, but to the first compositional field, and the last value was ignored. This is fixed now, the first entry is used for the background field, and all following values determine whether to include the compositional fields.
    (Rene Gassmoeller, 2018/05/21)

31. Fixed: The nullspace removal had a minor bug that caused a small remaining rotation in three dimensional models even if the net rotation or angular momentum should have been removed. The bug was fixed, and a new postprocessor called 'rotation statistics' was added that can be used to calculate net rotation and angular momentum.
    (Rene Gassmoeller, Shangxin Liu, 2018/05/03)

32. New: ASPECT now stores the steps for the continuous integration setup inside the repository in form of a Jenkins pipeline file. This file documents each step in our testing procedure, and allows to easily setup new Jenkins server that reproduce the official ASPECT tester.
    (Rene Gassmoeller, Timo Heister, Tyler Esser, 2018/05/02)