Financial planning problems are formulated as large scale, stochastic, multi-period, tree structured optimization problems. An efficient technique for solving this kind of problems is the nested Benders decomposition method. In this paper we apply this technique to the problem of portfolio optimization and present a parallel, portable, asynchronous implementation. To achieve our portability goals we elected the programming language Java for our implementation and used a high level Java based framework, called OpusJava, for expressing the parallelism potential as well as synchronization constraints. Our implementation is embedded within a modular decision support tool for portfolio and asset liability management, the Aurora Financial Management System.

Top