- easy-to-catch concept of GMsFEM shown in the NumPy implementation
- modular structure that allows targeted assembling
- convinient API for FEniCS-2019 docker containers
- parallel pipeline (not fully completed)
-
it includes all drawbacks connected with FEniCS-2019.1, like bugs or unimplemented features. Major releases (e.g. dolfinx) coming after make this implementation a bit outdated
-
the code is not fully-optimized. For example, it can be rewritten in C++ which will half at least the the runtime of some of its sections. A few algorithmic features may be improved as well
- Write a C++ implementation
- Use DOLFINX (when it is released)
- Substitute triangular coarse elements for square ones
- Finish README
- Add MIT license