SNOPT is a widely used large-scale optimizer for difficult large-scale nonlinear problems. It incorporates proven methods that have wide applicability and are especially effective for nonlinear problems whose functions and gradients are expensive to evaluate.
Developer: Stanford Systems Optimization Laboratory
Current version: 7.2
Problem types supported: Linear, quadratic, and smooth nonlinear objectives and constraints in continuous variables.
Algorithms available: Primal simplex for linear problems; sequential quadratic with augmented Lagrangian for nonlinear objectives and constraints.
Special features: Particularly great efficiency is achieved when many variables enter linearly in the constraints, or when many constraints are active (hence there are relatively few degrees of freedom) at an optimum. An elastic constraint formulation is used to deal with infeasibility.
Systems Optimization Laboratory website
SNOPT for AMPL option listing
SNOPT 7 User’s Guide with complete option descriptions