Abstract:In an effort to develop an alternative approach to traditional sparse reformulations, we will provide a new type of convex reformulation of a large class of stochastic quadratically constrained quadratic optimization problems that is similar to Burer's reformulation, but lifts the variables into a comparatively lower dimensional space. The reformulation rests on a generalization of the set-completely positive matrix cone. This cone can then be approximated via inner and outer approximations in order to obtain upper and lower bounds, which potentially close the optimality gap, and hence can give a certificate of exactness for these sparse reformulations outside of traditional, known sufficient conditions. Finally, we provide some numerical experiments, where we asses the quality of the inner and outer approximations, thereby showing that the approximations may indeed close the optimality gap in interesting cases.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively perform sparse conic relaxation on Structured Quadratically Constrained Quadratic Programs (QCQPs) in stochastic optimization, especially methods based on copositive optimization. Specifically, the paper focuses on how to maintain the quality of solutions while significantly reducing computational complexity through sparse conic relaxation methods when dealing with two - stage stochastic optimization problems. ### Background and Motivation In traditional copositive optimization methods, the dimension of the variable space will increase sharply as the number of scenarios \( S \) increases, which makes the optimization problem very difficult to solve. To solve this problem, Bomze et al. proposed a sparse copositive relaxation method, which requires only \( O(S(n_1 + n_2)^2) \) variables instead of \( O((n_1 + Sn_2)^2) \) variables. Although this method is more computationally efficient, the quality of its solutions is almost the same as that of traditional methods, and in some cases there is almost no difference. This phenomenon lacks a theoretical explanation and becomes the main motivation for this research. ### Main Contributions 1. **New Sparse Conic Relaxation Method**: - The author introduced a new sparse conic relaxation method based on a generalized set - completely positive matrix cone. This method is not only more computationally efficient but also can provide solution quality comparable to traditional methods. - Through inner approximation and outer approximation, the author provided upper and lower bounds, which can be used to verify the optimality of the sparse relaxation method. 2. **Theoretical Analysis**: - The author discussed in detail why existing sparse conic optimization methods cannot explain this phenomenon and proposed a new theoretical framework to explain the effectiveness of the sparse relaxation method. - The author proved that under specific conditions, the solution quality of the sparse relaxation method is comparable to that of traditional methods. 3. **Numerical Experiments**: - The author verified the effectiveness of the new method through numerical experiments. The results show that in many cases, the new method can significantly reduce the computation time while maintaining a high solution quality. ### Method Overview 1. **Problem Definition**: - The paper first defined a class of generalized quadratic optimization problems, which cover the standard quadratic optimization problems in two - stage stochastic optimization. 2. **Sparse Conic Relaxation**: - The author proposed a new sparse conic relaxation method, which reduces the computational complexity by lifting the variable space to a lower - dimensional space. - This method is based on a generalized set - completely positive matrix cone and obtains upper and lower bounds through inner approximation and outer approximation. 3. **Theoretical Analysis**: - The author analyzed in detail the effectiveness of the sparse relaxation method and proved that under specific conditions, the solution quality of the sparse relaxation method is comparable to that of traditional methods. 4. **Numerical Experiments**: - The author verified the effectiveness of the new method through a series of numerical experiments. The results show that the new method can significantly reduce the computation time while maintaining a high solution quality in many cases. ### Conclusion This paper solves the problem of how to effectively handle Structured Quadratically Constrained Quadratic Programs in two - stage stochastic optimization by introducing a new sparse conic relaxation method. The new method is not only more computationally efficient but also can maintain solution quality comparable to traditional methods. Through theoretical analysis and numerical experiments, the author provides a solid theoretical basis for the effectiveness of the sparse relaxation method.

Sparse Conic Reformulation of Structured QCQPs based on Copositive Optimization with Applications in Stochastic Optimization

A GLOBALLY AND SUPERLIEARLY CONVERGENT FEASIBLE SQP ALGORITHM FOR DEGENERATE CONSTRAINED OPTIMIZATION

Finding quadratic underestimators for optimal value functions of nonconvex all-quadratic problems via copositive optimization

Conic Approximation to Quadratic Optimization with Linear Complementarity Constraints.

Simplex QP-based methods for minimizing a conic quadratic objective over polyhedra

Tighter yet more tractable relaxations and nontrivial instance generation for sparse standard quadratic optimization

On Tractable Convex Relaxations of Standard Quadratic Optimization Problems under Sparsity Constraints

Conic approximation to nonconvex quadratic programming with convex quadratic constraints

Quadratic unconstrained binary optimization problem preprocessing: Theory and empirical analysis

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

Further Development in Convex Conic Reformulation of Geometric Nonconvex Conic Optimization Problems

The sparse(st) optimization problem: reformulations, optimality, stationarity, and numerical results

Adaptive computable approximation to cones of nonnegative quadratic functions

Tightening a Copositive Relaxation for Standard Quadratic Optimization Problems

Exploiting cone approximations in an augmented Lagrangian method for conic optimization

Uncertain standard quadratic optimization under distributional assumptions: a chance-constrained epigraphic approach

Concave tents: a new tool for constructing concave reformulations of a large class of nonconvex optimization problems

A Globally Convergent Stabilized SQP Method

A Matrix-Free Trust-Region SQP Method for Equality Constrained Optimization

An efficient nonconvex reformulation of stagewise convex optimization problems

Second-Order Conic and Polyhedral Approximations of the Exponential Cone: Application to Mixed-Integer Exponential Conic Programs