Long-term goals

High priority items are marked with a "🔥" symbol.

Charisee is working on items with a 🌹 symbol.

##Compiler Development

differentiation 🔥
inside test 🔥
- surface to high-to-mid 🌹
create a richer support for elements
surface syntax clean up
- create tuple? for description and pointer
EIN params should reflect mesh/element/space
Tighten type checker for different input
- Do we still need three pointers? Teo: No.
code generation optimization
- (need benchmarks)
- some clean up should help
investigate locality issues with regard to mesh
Push to mid (discuss as it might make differentiation easier)
merge
- backend in sync 🌹
- flux with chiw branch
Do some benchmarks
Optimizations: ** Parallelization -> look at generated code examples- > Charisee setups an ** tracking by cell ** trying to predict the next cell ** asking firedrake for a nice numbering system ** liobspatial index

##Examples

crest lines or some vis feature with high level of tensor math (needs differentiation)
higher order vs. lower order
Demonstrate need by showing improvement on examples that paraview cannot visualize well. ** Teo has a question about plotting things and is putting this here for memory. ** When is higher degree but smaller mesh cheaper?

##testing

regression tests for point-eval: Teo: There are a lot of tests being run via testing/test.py as of this moment.
visualization correctness
- symmetry
list of pdes with known solutions that FEM can solve: Evans and Andrei D. Polyanin, Vladimir E. Nazaikinskii Handbook of Linear Partial Differential Equations for Engineers and Scientists (I have PDFs)
talk about metamorphic testing

Writing