Oscar Dowson

@conema if we want to re-register this as a package, "DynamicProgramming.jl" is likely to cause some issues because it's a little too generic. This isn't _THE_ package for dynamic programming...

These would be nice to have. ```julia @states(sp, begin x in 0:10 y in [:a, :b, :c] end) ```

Unify interface with SDDP.jl

1

Ref https://github.com/odow/SDDP.jl

Particularly those from Bertsekas, D. (2005). _Dynamic Programming and Optimal Control_ (3rd ed.). Bellmont, MA: Athena Scientific. Bellman, R., & Dreyfus, S. (1962). _Applied Dynamic Programming_. Princeton, NJ: Princeton University...

- [ ] parallel code - [ ] maximization examples

Explore repeated terms in a function

8

A lot of problems have the form: `sum_i f(x_i)`, for example machine learning problems where the sum is over observations in a dataset. We can improve the performance of this...

Implement `RelativeGap`

3

Instead of passing off to the solver https://github.com/JuliaOpt/LinQuadOptInterface.jl/blob/812d334cc26394f4409c782ebdf6b4cd16e9debe/src/solve.jl#L244-L246 we should implement this ourselves to avoid issues such as https://github.com/JuliaOpt/MathOptInterface.jl/issues/590.

Implement `copy_to`

5

An efficient implementation of `MOI.copy_to` is needed when JuMP is used in `AUTOMATIC` mode. See https://github.com/JuliaOpt/Gurobi.jl/pull/188#issuecomment-459385303.

Farkas duals for variable bounds

4

Currently we only get the farkas dual for constraints: https://github.com/JuliaOpt/LinQuadOptInterface.jl/blob/d6957f357227e692c0e15bd6c64e9d0a37b7b1f1/src/solve.jl#L58 However, `linear12` needs it for variable bounds.