- Rewriting local kronmult library and adding an optional global kronmult library
- Matrix-free implicit timestepping using GMRES or BICGStab solver
- IMEX time advance
- File checkpoint restart
- New PDEs
- multi-node capability (merged 24 Oct 2019)
- enable optional MPI build dependency
- performance improvements
- batching (ptrs, parallel)
- tensors
- build improvements (cray, fortran)
- bugfixes
- enable profiling via CMake (merged 01 May 2019)
- single gpu capability for low-level code (merged 10 Sep 2019)
- CMake CUDA language capability (merged 03 Dec 2019)
- fk::tensors understand unattached memory
- blas on single gpu
- performance improvements
- forward transform
- bugfixes
- initial structure of the setup code
- permutations, connectivity component (merged 23 Jan 2019)
- transformations component (merged 08 Feb 2019)
- missing quadrature/utilities test inputs (merged 06 Feb 2019)
- updated PDE spec for arbitrary dimension (merged 13 Feb 2019)
- operator matrices (merged 06 Mar 2019)
- views for fk::tensors (merged 11 Apr 2019)
- batching for fk::tensors (views) (merged 17 May 2019)
- dimension agnostic kronecker products (merged 17 May 2019)
- time loop (merged 22 May 2019)