How to view intermediate steps in ODESolver during runtime without excess allocations?

New to Julia
12

I suspect that some of the type returned by the setup are not inferrable. Try a function barrier before you enter the loop and see whether this recovers the performance of solve!:

function inner(integrator)
    x = 0.0
    for i in integrator
        x = i.u 
    end
end

function integrate(alg=OrdinaryDiffEq.RK4())
    
    tspan = (0.0, 1e3)

    u0 = 1.0

    prob = ODEProblem{false}(diffeq, u0, tspan)
    integrator = OrdinaryDiffEq.init(prob, alg, save_everystep = false, maxiters = 1e4, abstol=1e-8, reltol=1e-8) 
    inner(integrator)
end 

@btime integrate()
2 Likes