当前位置: X-MOL 学术Optim. Lett. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Structured linear reformulation of binary quadratically constrained quadratic programs
Optimization Letters ( IF 1.6 ) Pub Date : 2018-11-17 , DOI: 10.1007/s11590-018-1361-8
Shan Jiang , Shu-Cherng Fang , Tiantian Nie , Qi An

This paper presents a new linear reformulation to convert a binary quadratically constrained quadratic program into a 0–1 mixed integer linear program. By exploiting the symmetric structure of the quadratic terms embedded in the objective and constraint functions, the proposed linear reformulation requires fewer variables and constraints than other known O(n)-sized linear reformulations. Theoretical proof shows the proposed reformulation provides a tighter linearization for each quadratic term comparing to other known linear reformulations. Extensive numerical experiments support the superior computational efficiency of the proposed reformulation in terms of the running time and number of nodes explored.

中文翻译:

二进制二次约束二次程序的结构化线性重构

本文提出了一种新的线性公式,它将二进制二次约束二次程序转换为0-1混合整数线性程序。通过利用嵌入在目标函数和约束函数中的二次项的对称结构,与其他已知的On)大小的线性重构相比,提出的线性重构需要更少的变量和约束。理论证明表明,与其他已知的线性重构相比,所提出的重构为每个二次项提供了更严格的线性化。大量的数值实验在运行时间和所探索的节点数量方面都支持拟议的公式的卓越计算效率。
更新日期:2018-11-17
down
wechat
bug