Dear all,

This question is related to the DifferentialEquations.jl package.

I’ve used the very helpful documentation here to implement adjoint sensitivity analysis on an ODE. Specifically, i have a cost function of the form

C(\theta) = \int^T_0 L[x(t)] \ dt ,

where x(t) is the state of my ODE over time. Calling the function adjoint_sensitivities gives me \nabla C(\theta, the gradient of the cost in the parameters.

Is there some callback I can use that allows me to get the actual cost function C(\theta) itself? Currently I have to have a separate routine for getting the cost, and getting the cost gradient

Thanks