LOQO: An Interior Point Code for Quadratic Programming

Robert Vanderbei

This paper describes a software package, called LOQO, which implements a primal-dual interior-point method for linear and quadratic programming. We describe in detail the algorithm emphasizing, in particular, the benefits of constructing primal-dual interior-point methods that preserve primal-dual symmetry. We show that the industry standard MPS format fits nicely into the symmetric formulation paradigm. Computational results are included showing, among other things, that LOQO solves the entire NETLIB suite of linear programs 2.25 times faster than CPlex-barrier and is 250 times faster on at least one problem in this suite.

Technical Report SOR 94--15, Program in Statistics & Operations Research, Princeton University, February 1995.

Contact: rvdb@princeton.edu