A Julia Package for Solving Deterministic Global Optimization Problems with Differential Constraints
EAGO_Differential.jl is a work in progress and is currently not a registered Julia package. As such, it should be added or developed using the package url as detailed in the Pkg documentation. It has been tested using the this commit of the EAGO master branch.
The package uses the EAGO architecture to formulate parametric ODE constrained optimization problems as a series of block-sequential implicit function solution routines and constructs the relaxation using the theory presented in [1,2,3] and a forthcoming paper. This package exports 'solve_ode' function used to solve parametric ODEs of the above form as well as 'ImplicitODELowerEvaluator' and 'ImplicitODEUpperEvaluator' structures which are used to calculated the convex/concave relaxations and associated interval bounds used in the optimization algorithm.
- Variable step-size algorithms.
- Solving parametric ODE constrained semi-infinite programs.
- Additional implicit approaches and mixed explicit-implicit approaches.
Work contained in this package and associated theory has been submitted for peer review. For now, please cite this package as software.
- Stuber, M.D., Scott, J.K., and P.I. Barton. Convex and Concave Relaxations of Implicit Functions. Optimization Methods and Software. 30(3), 424-460, 2014.
- Scott, J.K., Stuber, M.D., and Barton, P.I. Generalized McCormick Relaxations. J Global Optim, 51:569-606, 2011