Aditya Sengupta

That works when all the data are the same type, but I've got several interpolations all of different types because they're carrying Unitful.jl units (I'm dealing with an atmosphere so...

Yes, these are quantities that are usually specified through predefined data arrays at discrete sample points, but I want to treat them as continuous functions without rebuilding the function at...

Users might request values beyond the range, in which case I'd like the Flat boundary condition to be used without having to specify, e.g. y(max(x, x1)) On Wed, Apr 20,...

I think my use case was a better fit to DataInterpolations.jl.

Possibly unrelated points, but I wasn't sure whether they were worth opening a new issue: - do SampledData blocks support non-uniform sampling? As far as I can tell they just...

> They could, but the ODE solver only ever calls them at a single time point at a time so I'm not sure what the optimization would be here? Makes...

Sure - I have sampled data that I'd like to feed into source terms for an ODE. The (data) -> (source term) transformation is a function that I can `@register_symbolic`,...

I changed the example you linked a bit to generate my MWE. ``` using ModelingToolkit using ModelingToolkitStandardLibrary.Blocks: SampledData, Parameter using OrdinaryDiffEq using Plots using DataInterpolations @parameters t D = Differential(t)...

Undoing some of my changes in accordance with https://github.com/SciML/ModelingToolkitStandardLibrary.jl/issues/253 so I can just use `@named` instead of passing around a kwarg and get closer to the expected behaviour. The substantial...

Found a workaround, but I don't know if this is idiomatic for MTK; it uses some awkward type mixing. ``` @named system = System() sys = structural_simplify(system) prob = ODEProblem(sys,...