Optimal Control Problems implemented in Snopt within MatLab
Contained within this directory are various Optimal Control problems numerically solved using the MatLab interface to the nonlinear solver SNOPT.
These projects will run on the student/trial versions of SNOPT, which can be obtained at http://www.scicomp.ucsd.edu/~peg/Software.html
Projects contained within this directory:
- Brachistochrone: Here Snopt is used to numerically solve the brachistochrone problem under the influence of a resistive force proportional to the velocity of the particle. Various implementations are given, namely showcasing the explicitly computation of the Jacobian (of the decision variables with respect to the constraints) to increase convergence to the optimal solution, along with the computation of the sparsity pattern of the Jacobian. Similarly, a version where Snopt calculates the Jacobian is given; comparing the two shows the order-of- magnitude in run-time. Included in each is a feasibility analysis to the solution Snopt returns. Feasibility Analysis is a critical component to verifying and validating that the Numerical Solver (in this case SNOPT) indeed has found an optimal solution.