Algorithm Gurobi uses to solve general convex optimization

Answered

Comments

1 comment

  • Jaromił Najman

    Hi Miranda,

    Gurobi uses an interior point algorithm with predictor and corrector steps and a barrier objective function. The algorithm terminates when the difference between the primal and dual objective value reaches some given threshold. For QCPs, this threshold is defined by BarQCPConvTol and by BarConvTol for LPs.

    For LPs, the Crossover algorithm is used to obtain an extreme point as a solution point. This is not possible for QCPs due to the nonlinear constraints, which may force the optimal solution point away from an extreme point of the polyhedron defined by the remaining linear constraints.

    Best regards,
    Jaromił

    1
    Comment actions Permalink

Please sign in to leave a comment.

Powered by Zendesk