Compact mixed-integer programming formulations in quadratic optimization

被引:3
作者
Beach, Benjamin [1 ]
Hildebrand, Robert [1 ]
Huchette, Joey [2 ]
机构
[1] Virginia Tech, Grad Dept Ind & Syst Engn, Blacksburg, VA 24061 USA
[2] Rice Univ, Dept Computat & Appl Math, Houston, TX USA
关键词
Quadratic optimization; Nonconvex optimization; Mixed-integer programming; Gray Code; DIFFERENTIABLE CONSTRAINED NLPS; GLOBAL OPTIMIZATION; ALPHA-BB; PERSPECTIVE CUTS; MINIMIZATION; MODELS;
D O I
10.1007/s10898-022-01184-6
中图分类号
C93 [管理学]; O22 [运筹学];
学科分类号
070105 ; 12 ; 1201 ; 1202 ; 120202 ;
摘要
We present a technique for producing valid dual bounds for nonconvex quadratic optimization problems. The approach leverages an elegant piecewise linear approximation for univariate quadratic functions due to Yarotsky (Neural Netw 94:103-114, 2017), formulating this (simple) approximation using mixed-integer programming (MIP). Notably, the number of constraints, binary variables, and auxiliary continuous variables used in this formulation grows logarithmically in the approximation error. Combining this with a diagonal perturbation technique to convert a nonseparable quadratic function into a separable one, we present a mixed-integer convex quadratic relaxation for nonconvex quadratic optimization problems. We study the strength (or sharpness) of our formulation and the tightness of its approximation. Further, we show that our formulation represents feasible points via a Gray code. We close with computational results on problems with quadratic objectives and/or constraints, showing that our proposed method (i) across the board outperforms existing MIP relaxations from the literature, and (ii) on hard instances produces better bounds than exact solvers within a fixed time budget.
引用
收藏
页码:869 / 912
页数:44
相关论文
共 55 条
  • [51] Mixed-Integer Models for Nonseparable Piecewise-Linear Optimization: Unifying Framework and Extensions
    Vielma, Juan Pablo
    Ahmed, Shabbir
    Nemhauser, George
    [J]. OPERATIONS RESEARCH, 2010, 58 (02) : 303 - 315
  • [52] Triangular function analysis
    Wei, YC
    [J]. COMPUTERS & MATHEMATICS WITH APPLICATIONS, 1999, 37 (06) : 37 - 56
  • [53] Wiese S., 2021, COMPUTATIONAL PRACTI
  • [54] Globally Solving Nonconvex Quadratic Programs via Linear Integer Programming Techniques
    Xia, Wei
    Vera, Juan
    Zuluaga, Luis F.
    [J]. INFORMS JOURNAL ON COMPUTING, 2020, 32 (01) : 40 - 56
  • [55] Error bounds for approximations with deep ReLU networks
    Yarotsky, Dmitry
    [J]. NEURAL NETWORKS, 2017, 94 : 103 - 114