search

SciML / OrdinaryDiffEq.jl

High performance ordinary differential equation (ODE) and differential-algebraic equation (DAE) solvers, including neural ordinary differential equations (neural ODEs) and scientific machine learning (SciML)
https://diffeq.sciml.ai/latest/
Other
562 stars 211 forks source link

`autodiff = Val{true}()` vs `autodiff = true` in solver definitions #2391

Open ArnoStrouwen opened 3 months ago

ArnoStrouwen commented 3 months ago

Some solvers use Val, others not: https://github.com/SciML/OrdinaryDiffEq.jl/blob/master/lib/OrdinaryDiffEqRosenbrock/src/algorithms.jl#L114 https://github.com/SciML/OrdinaryDiffEq.jl/blob/master/lib/OrdinaryDiffEqRosenbrock/src/algorithms.jl#L155

Is this intentional, or should we standardize to one of the two?

ChrisRackauckas commented 3 months ago

We should standardize on Val.

But I wouldn't worry about it. We're changing this autodiff argument to use ADTypes.jl hopefully in the next month. That's a bigger breaking change.