Mission and General Plan
The goal of ASPECT is to provide the geoscience community with an extensible software written in C++ to support research in simulating convection in the Earth mantle and elsewhere by providing a well-documented, tested code base.
The general goals are:
- To include community developed features: provide help and review contributions
- To run tutorials and hackathons
- To fix bugs and maintain the code base
- Benchmarking and implementation of new features based on CIG and community feedback
- To provide regular software releases
- To provide support via mailing list, github issues, etc.
Suggested current work items (2017)
- Interface to couple BurnMan & ASPECT
- Revision of the 2nd ASPECT paper
- Work on deal.II related features
- Initial work on coupling with the mineral physics toolbox BurnMan
- Deprecation of outdated features towards ASPECT 2.0
- Redesign of non-linear solver infrastructure (tests/benchmarks)
- Merge the Newton solver
- Implement and test the new parameter GUI
Long term plan
- Stokes Solver improvements (melt preconditioner, Schur complement improvements, GMG)
- Benchmarking of different compressible formulations
- Participate in community benchmark efforts
- Perform parallel performance benchmarking of deal.II and ASPECT to increase efficiency
- Develop a robust non-linear solver framework
- Develop a robust and scalable passive and active tracer code
- Improve interoperability with codes used in other fields of study e.g. mineral physics, seismology, or the planetary sciences
- Provide coupling with the mineral physics toolbox BurnMan
Completed Items
Early 2017:
- ASPECT 1.5.0 release
- We successfully benchmarked ASPECT with the Blankenbach benchmarks
- Initial work on parameter GUI
Oct 2016 – Jan 2017:
- Mini hackathon in December before AGU
- The 2nd ASPECT paper got submitted
- We implemented and merged a correct Boussinesq, ALA, and TALA approximation
- We successfully benchmarked ASPECT with the TanGurnis and King2010 benchmarks
- The melt paper has been published
- The free surface paper “Stability and accuracy of free surface time integration in viscous flows” has been accepted
- We have merged a large number of improvements to the particle code, making it vastly faster than it was before
- A paper describing the techniques underlying the particle code has been written and submitted
- We wrote an initial BurnMan coupling module for adiabatic conditions coming out of mineral physics data