Benoît Legat
Benoît Legat
If this was only about `moi_function` then I would just get the model from the first `JuMP.VariableRef` I would see and then use the dictionary inside it. The reason I...
It would be weird to have the subexpressions only work when they are on the same function. Especially since the AD backend supports them being on different ones. I guess...
That makes sense, if we only take care of not duplicating aliased subexpression at the level of each function, it's indeed easier. We then detect subexpressions in the AD backend...
Thanks ! I would suggest a mix of the first and third option (but without the weakdeps). The methods compiled are of two types: ## JuMP methods Methods with `JuMP.Model`...
We could add support for it in this bridge but many other bridges don't support it. So we should maybe add a cache, it could be an option of POI....
Yes and in case we have this error, we should just tell the user to use the option that adds a cache
> If the manifolds require ManifoldDiff or user-defined operators (like `distance`), then the answer is a strong "no", this cannot be done in JuMP. The main reason is that we...
> if one wants to compute large numbers of Jacobian-Vector products, it may be more efficient to form the full Jacobian `J`, then do matrix-vector products. I think the code...
We could maybe consider using https://github.com/julia-vscode/TestItems.jl
I'll first make https://github.com/jump-dev/MatrixOptInterface.jl/pull/35 and then take a miniature version of https://github.com/jump-dev/MatrixOptInterface.jl/pull/35 as test here