This jupiter notebook supplements the written report on the adjoint state method. It implements
- The difference of computational in "forward" and "backward" matrix - vector products
- A nonlinear optimization problem solved with an adjoint method and gradient descent
- A comprehensive clean code example of the adjoint state method that only uses ODE solvers. Comparison to finite differences
For a computationally efficint sensitivity calculation for ODEs and other equations, see the SciMLSensitivity.jl package for julia.