Solving for the Global Nonlinear Saddlepath: Reverse Shooting vs. Approximation Methods

The paper implements reverse shooting to solve for global nonlinear saddle path for optimal control problems with two state variables. It shows how to do reverse shooting. This will reduce entry barriers for researchers for whom this technique may be useful. For this purpose, user-friendly ‘plug-and- play’ programs are made available in public domain. These sample programs can be easily tailored to solve other problems as the search for the desired saddle path is done with very limited use of problem-specific information. The paper also compares the performance of existing approximation methods and finds their ranking in general depends on the nature of the problem, the variables that are of interest and the time horizon of interest