M. Ayoub Chettouh

New Adapter: Julia

2

Hello, I'd like to create a Julia Test adapter. I'm kinda new so some help would be nice. Running tests in Julia is simple for the most part, I'll try...

Interpolating if one or more of the dimensions has length 1.

2

#304 Still has a warning assertion in case all ivs have length 1. Otherwise it will build an interpolation object for the non-"singleton" ivs. If the model is prompted to...

Zygote not working for gradient w.r.t. parameters (with remake)

15

MWE: ````julia using DifferentialEquations, ModelingToolkit, MethodOfLines, DomainSets using Zygote import AbstractDifferentiation as AD # Method of Manufactured Solutions: exact solution u_exact = (x,t) -> exp.(-t) * cos.(x) # Parameters, variables,...

````julia using Preferences, UUIDs set_preferences!(UUID("0bca4576-84f4-4d90-8ffe-ffa030f20462"), "SpecializationLevel" => "FullSpecialize") ```` doesn't change the type of the ODE problem.

Possibility to restart solving a system starting from a solution (ics override)

6

Basically the possibility to do `solve(discretized_problem, previous_solution, p = new_parameters, otherkwargs...)` by overriding the previous ics (or in some other way).

Interpolation in space not possible if solution is only saved at the end.

4

If the solution is constrained to the last point in time, interpolation will no be possible. ````julia using DifferentialEquations, ModelingToolkit, MethodOfLines, DomainSets # Method of Manufactured Solutions: exact solution u_exact...

Generalized save

2

#309 Still needs some docs

`savedataset` has a `path` kwarg with a default value of `""`, wouldn't this give unexpected behavior? Shouldn't `path` have no default value (or better yet be a non-kwarg)? @meggart

Inconsistency in syntax between `savecube` and `savedataset`

2

Why is the path `savecube` a "normal" argument, while path in `savedataset` a kwarg?

It would be nice to implement a predictive chunk loader akin to branch prediction and cache loading.