Benoît Legat
Benoît Legat
In https://discourse.julialang.org/t/differentiating-through-a-qp-order-of-constraints/119275/3, it is argued that the input of the differentiation should be reset in-between different calls to `DiffOpt.forward_differentiate!(model)`. We should clarify this in the documentation.
Before ``` BenchmarkTools.Trial: 10000 samples with 4 evaluations. Range (min … max): 8.011 μs … 40.041 ms ┊ GC (min … max): 0.00% … 99.93% Time (median): 8.852 μs ┊...
This is useful in constraint programming where you often have a set that is a special case of another one. One example is CBLS.jl that has sets with a few...
We recommend to only support a constraint if the solver has a native support for it (see this tip: https://jump.dev/JuMP.jl/stable/moi/tutorials/implementing/#Define-supports_constraint). The reason is that the bridging mechanism in MOI selects...
@kalmarek I'll wait for KnuthBendix to integrate `Monoids` before merging this, I'll leave it as a WIP for now. I needed a workaround with `Full` for it to work with...
This was a draft a did some time ago. The goal now is to wait for https://github.com/kalmarek/SymbolicWedderburn.jl/issues/79
It would be nice if `dualize` could optionally add a bridge layer before dualizing so that it can for instance reformulate quadratic constraints to SOC