Abstract:Decades of advances in mixed-integer linear programming (MILP) and recent development in mixed-integer second-order-cone programming (MISOCP) have translated very mildly to progresses in global solving nonconvex mixed-integer quadratically constrained programs (MIQCP). In this paper we propose a new approach, namely Compact Disjunctive Approximation (CDA), to approximate nonconvex MIQCP to arbitrary precision by \textit{convex} MIQCPs, which can be solved by MISOCP solvers. For nonconvex MIQCP with $n$ variables and $m$ general quadratic constraints, our method yields relaxations with at most $O(n\log(1/\varepsilon))$ number of continuous/binary variables and linear constraints, together with $m$ \textit{convex} quadratic constraints, where $\varepsilon$ is the approximation accuracy. The main novelty of our method lies in a very compact lifted mixed-integer formulation for approximating the (scalar) square function. This is derived by first embedding the square function into the boundary of a three-dimensional second-order cone, and then exploiting rotational symmetry in a similar way as in the construction of BenTal-Nemirovski approximation. We further show that this lifted formulation characterize the union of finite number of simple convex sets, which naturally relax the square function in a piecewise manner with properly placed knots. We implement (with JuMP) a simple adaptive refinement algorithm. Numerical experiments on synthetic instances used in the literature show that our prototypical implementation (with hundreds of lines of Julia code) can already close a significant portion of gap left by various state-of-the-art global solvers on more difficult instances, indicating strong promises of our proposed approach.

Conic approximation to nonconvex quadratic programming with convex quadratic constraints

Conic Approximation to Quadratic Optimization with Linear Complementarity Constraints.

Adaptive computable approximation to cones of nonnegative quadratic functions

Convex Relaxations for Nonconvex Quadratically Constrained Quadratic Programming: Matrix Cone Decomposition and Polyhedral Approximation

Exploiting cone approximations in an augmented Lagrangian method for conic optimization

Quadratic Optimization over A Polyhedral Cone

A Conic Approximation Method for the 0-1 Quadratic Knapsack Problem

Quadratic Optimization over One First-Order Cone

Quadratic Optimization over a Second-Order Cone with Linear Equality Constraints

A Perturbation Approach for an Inverse Quadratic Programming Problem

Quadratic Convergence of a Smoothing Newton Method for Symmetric Cone Programming Without Strict Complementarity

Extended canonical duality and conic programming for solving 0-1 quadratic programming problems

Nonlinearly Constrained Best Approximation in Hilbert Spaces: the Strong CHIP and the Basic Constraint Qualification.

Compact Disjunctive Approximations to Nonconvex Quadratically Constrained Programs

The Augmented Lagrangian Method for a Type of Inverse Quadratic Programming Problems over Second-Order Cones

Computable Representation of the Cone of Nonnegative Quadratic Forms over A General Second-Order Cone and Its Application to Completely Positive Programming

A Sequential Quadratically Constrained Quadratic Programming Method of Feasible Directions

Improved Approximation Results on Standard Quartic Polynomial Optimization

Second Order Cone Constrained Convex Relaxations for Nonconvex Quadratically Constrained Quadratic Programming

Kkt Solution And Conic Relaxation For Solving Quadratically Constrained Quadratic Programming Problems

Sub-Quadratic Convergence of A Smoothing Newton Method for Second-Order Cone Programming