SQPpro - A solver of nonlinear optimization problems, using an SQP approach

Jean Charles Gilbert (1)

1 ESTIME - Parameter estimation and modeling in heterogeneous media

SQPpro is a piece of software that aims at solving a nonlinear optimization problem with nonlinear equality and inequality constraints. The functions defining the problem must be at least once differentiable. The implemented algorithm uses an SQP approach, which is a workable version of the Newton and quasi-Newton methods. The quadratic optimization that has to be solved at each iteration uses the solver QPAL. The constraint Jacobian matrices can be strored in dense or sparse structures; in addition the Hessian of the Lagrangien can be approximated by the BFGS (dense) or $\ell$-BFGS (sparse) formula. SQPpro is written in Fortran-2003.

Mots clés: BFGS and l-BFGS updates; Newton and quasi-Newton methods; nonlinear optimization; QPAL quadratic optimization solver; SQP algorithm

Domaines: Optimisation et contrôle [math.OC]; Recherche opérationnelle [math.OC]

Jean Charles Gilbert. SQPpro - A solver of nonlinear optimization problems, using an SQP approach. [Technical Report] RT-0378, INRIA. 2009, pp.24. ⟨inria-00442314⟩