Linear Interpolation as an algebraic function in a reaction network

[danilodurs]

Hello ,

I am currently using Catalyst to build up a reaction network from which I generate a system of ODEs .
The model I am implementing implies interpolated data as an algebraic function, as described in Bergman's minimal model (Bergman et al 1979) :

Iintpol(t) = linear_interpolation(tticks,datamean2)(t)
minimalmodel = @reaction_network minimalmodel begin
    B_0,     ∅   ⇒   G_p            # hepatic glucose production
    k_gelim, G_p --> ∅              # glucose utilization/clearance
    (k_iut + k_imet)*I_r, G_p --> ∅ # insulin-dependent glucose regulation
    k_gmet, G_p --> ∅               # glucose hepatic metabolization
    k_ipan*Iintpol(t), ∅ ⇒ I_r        # blood insulin --> remote insulin            
    k_ielim, I_r --> ∅              # insulin clearance
end B_0 k_gelim k_iut k_imet k_gmet k_ipan k_ielim

However, unlike other algebraic functions I have previously used, some implying the variable t, in different models, linear interpolation does not seem to work inside an ODE:

TypeError: non-boolean (Num) used in boolean context

I also have kind of the same problem when I directly implement the system of ODEs:

function fminimalmodel(u, p, t)
    B_0,k_gelim,k_iut,k_imet,k_gmet, k_ipan ,k_ielim = p
    G_p , I_r = u
    dGp = B_0 - k_gelim * G_p - k_gmet * G_p - (k_iut + k_imet) * G_p * I_r
    dIr = k_ipan * Iintpol(t) - k_ielim* I_r
    [dGp, dIr]
end
prm = [datamean1[1],0.03,0.04,0.005,0.09,0.9,0.005]
oprob = ODEProblem(fminimalmodel, [datamean1[1],0.0], tspan, prm)
sol = solve(oprob)

output:

BoundsError: attempt to access 9-element extrapolate(interpolate((::Vector{Union{Missing, Int64}},), ::Vector{Float64}, Gridded(Linear())), Throw()) with element type Float64 at index [305.0204580512005]

Is there any workaround in Catalyst or DifferentialEquations to solve this issue?

[isaacsas]

Can you post a complete minimum working example that loads the appropriate package and builds the interpolant so we can run the code ourselves? Your DifferentialEquations error seems to be related to the interpolant -- are you sure you are building / using it correctly? Have you checked you can call Iintpol(t) and plot it yourself from how you've defined it?

Regarding Catalyst, you probably need to register your function using @register_symbolic. See https://symbolics.juliasymbolics.org/stable/manual/functions/

[danilodurs]

For the DifferentialEquations object, it was just a misuse indeed (tspan[end] > tticks[end]): issue solved.
The register_symbolic macro did the trick, the model works as expected, I've never used it before, good to know.
Thank you