opened 09:56AM - 25 Sep 24 UTC
question
**Question❓**
Hi everyone,
I am working on an adapter for virtual commissi…oning using [ModelingToolkit.jl](https://github.com/SciML/ModelingToolkit.jl). More specifically, I use discrete callbacks (timed --> periodic) to read inputs from and write output to a soft-controller. Is there a way to set the `initial_affect` and the `final_affect` arguments from [DiffEqCallbacks.jl](https://github.com/SciML/DiffEqCallbacks.jl) using [ModelingToolkit.jl](https://github.com/SciML/ModelingToolkit.jl)? Here is my MWE:
```julia
using ModelingToolkit, DifferentialEquations
using ModelingToolkit: t_nounits as t, D_nounits as D
function affect!(integ, u, p, ctx)
println("Time: $(integ.t)")
end
@mtkmodel SimpleSystem begin
@parameters begin
m = 1.0
end
@variables begin
y(t) = 0.0
end
@equations begin
D(y) ~ m
end
@discrete_events begin
1 => (affect!,[],[],[],nothing)
end
end
@mtkbuild sys = SimpleSystem()
prob = ODEProblem(sys, Pair[], (0, 10.0))
sol = solve(prob)
```
If you run this code the event will not be triggered for t = 0.0 nor 10.0. If you explicitly add additional events like this:
```julia
@discrete_events begin
[0,10] => (affect!,[],[],[],nothing)
1 => (affect!,[],[],[],nothing)
end
```
both events will be called twise. I guess this is related to the solver starting at some $\Delta t$ in the first case (and thus overstepping 0.0) and being forced to start with 0.0 in the second case. Is there currently a way arround this behavior in [ModelingToolkit.jl](https://github.com/SciML/ModelingToolkit.jl)?