Abstract:Deterministic nonconvex optimization solvers generate convex relaxations of nonconvex functions by making use of underlying factorable representations. One approach introduces auxiliary variables assigned to each factor that lifts the problem into a higher-dimensional decision space. In contrast, a generalized McCormick relaxation approach offers the significant advantage of constructing relaxations in the lower dimensionality space of the original problem without introducing auxiliary variables, often referred to as a "reduced-space" approach. Recent contributions illustrated how additional nontrivial inequality constraints may be used in factorable programming to tighten relaxations of the ubiquitous bilinear term. In this work, we exploit an analogous representation of McCormick relaxations and factorable programming to formulate tighter relaxations in the original decision space. We develop the underlying theory to generate necessarily tighter reduced-space McCormick relaxations when a priori convex/concave relaxations are known for intermediate bilinear terms. We then show how these rules can be generalized within a McCormick relaxation scheme via three different approaches: the use of a McCormick relaxations coupled to affine arithmetic, the propagation of affine relaxations implied by subgradients, and an enumerative approach that directly uses relaxations of each factor. The developed approaches are benchmarked on a library of optimization problems using the EAGO.jl optimizer. Two case studies are also considered to demonstrate the developments: an application in advanced manufacturing to optimize supply chain quality metrics and a global dynamic optimization application for rigorous model validation of a kinetic mechanism. The presented subgradient method leads to an improvement in CPU time required to solve the considered problems to -global optimality.

Gaining or losing perspective for convex multivariate functions on a simplex

Gaining or losing perspective for convex multivariate functions on box domains

Convexification of box-constrained polynomial optimization problems via monomial patterns

Sparse convex relaxations in polynomial optimization

On generalized surrogate duality in mixed-integer nonlinear programming

Improved Convex and Concave Relaxations of Composite Bilinear Forms

Convexification, Concavification and Monotonization in Global Optimization

Convexification of Bilinear Terms over Network Polytopes

Convex Regression in Multidimensions: Suboptimality of Least Squares Estimators

An Alternative Perspective on Copositive and Convex Relaxations of Nonconvex Quadratic Programs

Disciplined Multi-Convex Programming

Convex optimization over a probability simplex

Convex Discrete Optimization

Constrained Optimization of Rank-One Functions with Indicator Variables

Relaxations and Duality for Multiobjective Integer Programming

The Multi-Objective Polynomial Optimization

A Slightly Lifted Convex Relaxation for Nonconvex Quadratic Programming with Ball Constraints

Non-convex mixed-integer nonlinear programming: A survey

Convexoid: A Minimal Theory of Conjugate Convexity