蒙特卡罗方法计算VAR的并行实现

Monte Carlo Method for Parallel Implementation of VAR

当今经济全球化、金融全球化进程加快,市场竞争愈演愈烈。建立一个良好的风险管理系统已成为各金融单位面临的重要问题。在众多的风险管理方法中最具代表性的是Value at Risk(VAR)方法。而计算VAR常见的三种方法中,Monte Carlo模拟最为有效。由于蒙特卡罗模拟需要大量数据,所以并行计算在计算VAR中非常重要。笔者基于Monte Carlo方法计算VAR的原理构造C++算法,并用基于API、Open MP、MPI三种多核并行算法,比较分析出了适合科学计算研究的并行算法。

Nowadays,the economic globalization and financial globalization are speeding up,and the market competition is becoming much fiercer.Establishing a good risk management system has become an important issue for financial institutions.Among the many risk management methods,the most representative is the Value at Risk(VAR)method.Among the three common methods for computing VAR,the Monte Carlo simulation is most effective.Since Monte Carlo simulation requires a large amount of data,parallel computing is very important in computing VAR.The author calculates the principle of VAR and constructs C++algorithm based on the Monte Carlo method,and compares parallel algorithms based on API,OpenMP,MPI and three multi-core parallel algorithms to compare the parallel algorithms which are suitable for scientific computing research.