摘要
结合金融机构层次组织结构,采用两层主从(m/w)计算模型机群系统,通过简化问题复杂性,建立银行信用风险CreditMetrics框架下VaR实时评估系统。同时,分析信用风险Monte Carlo仿真计算过程,给出减少串行计算成份,提高系统并行性的措施。通过具体实验,表明该优化算法能够明显提高系统加速效果。
Combining with the hierarchal organization,a credit risk value at risk （ VaR） real-time evaluating system based on CreditMetrics framework is designed by using two level master/ worker computing clusters. Meanwhile,the Monte Carlo simulation algorithm for the CreditMetrics framework VaR computing is analyzed. And a strategy to reduce the serial part of algorithm is put forward. Experience shows that it can improve speedup effectively.
出处
《科技管理研究》
北大核心
2010年第19期198-201,共4页
Science and Technology Management Research
基金
教育部哲学社会科学后期资助项目"计算金融探索与实践"(09JHQ014)
