摘要
本文研究金融优化中的离散单因素投资组合问题,该问题与传统投资组合模型的不同之处是决策变量为整数(交易手数),从而导致要求解一个二次整数规划问题．针对该模型的可分离性结构,我们提出了一种基于拉格朗日对偶和连续松弛的分枝定界算法。我们分别用美国股票市场的交易数据和随机产生的数据对算法进行了测试．数值结果表明该算法是有效的,可以求解多达150个风险证券的离散投资组合问题．
In this paper, we consider the discrete single-factor model in portfolio optimization. This model is of quadratic integer programs. The separable structure is exploited to derive lower bounds by Lagrangian decomposition scheme. A new branch-and-bound algorithm based on Lagrangian relaxation and continuous relaxation is proposed for this model. Extensive computational results are reported for test problems both from real- world stock market and randomly generated.
出处
《运筹学学报》
CSCD
北大核心
2006年第4期49-56,共8页
Operations Research Transactions
基金
Research supported by the National Natural Science Foundation of China under grants 10571116 and 70518001.
关键词
运筹学
金融优化
离散单因素模型
拉格朗日松弛和连续松弛
分枝定界法
Operation research, portfolio optimization, discrete single-factor model, Lagrangian relaxation and continuous relaxation, branch-and-bound method
