PROPT is built on top of the source transformation package TomSym in the TOMLAB Base Module.
Optimally coded analytical first and second order derivatives, including problem sparsity patterns are automatically generated, thereby making it the first Matlab package to be able to fully utilize NLP (and QP) solvers such as: KNITRO, CONOPT, SNOPT and CPLEX, which may require second order information.
PROPT currently uses Gauss or Chebyshev-point collocation for solving optimal control problems. However, the code is written in a more general way, allowing for a DAE (Differential Algebraic Equations) rather than an ODE (Ordinary Differential Equations) formulation. Parameter estimation problems are also possible to solve.
Automatic scaling of difficult space problems is possible as well as the use of binary or integer states, control variables or parameters.