1: \begin{abstract}

2: We propose a new method for linear second-order cone programs.  

3: %

4: It is based on the sequential quadratic programming framework for nonlinear programming. In contrast to interior point methods, it can capitalize on the warm-start capabilities of active-set quadratic programming subproblem solvers and achieve a local quadratic rate of convergence.

5: %

6: 

7: In order to overcome the non-differentiability or singularity observed in nonlinear formulations of the conic constraints, the subproblems approximate the cones with polyhedral outer approximations that are refined throughout the iterations.

8: %

9: For nondegenerate instances, the algorithm implicitly identifies the set of cones for which the optimal solution lies at the extreme points.  As a consequence, the final steps are identical to regular sequential quadratic programming steps for a differentiable nonlinear optimization problem, yielding local quadratic convergence.

10: 

11: 

12: We prove the global and local convergence guarantees of the method and present numerical experiments that confirm that the method can take advantage of good starting points and can achieve higher accuracy compared to a state-of-the-art interior point solver.

13: 

14: %In order to capture the extreme points of the cones at which the nonlinear formulation of the conic constraint is not differentiable, the subproblems include polyhedral outer approximations of the cones that are improved throughout the iterations.

15: %

16: %A key feature of the algorithm is that this makes it possible to implicitly identify the set of cones for which the optimal solution lies at the extreme points, assuming nondegeneracy and strict complementarity.

17: %

18: %As a consequence, the final steps are identical to regular sequential quadratic programming steps for a differentiable nonlinear optimization problem and yield rapid local convergence.

19: 

20: 

21: \end{abstract}

22: