Parallel computing in asian option pricing

Abstract

We discuss the use of parallel computing in Asian option pricing and evaluate the efficiency of various algorithms. We only focus on "backward-starting fixed strike" continuously averaged Asian options. We implement a one-state-variable partial differential equation (P.D.E.) approach (Rogers and Shi (1995) and Alziary et al. (1997)) to price an Asian option. We also implement the same methodology to price a standard European option as the accuracy check for Asian option. We solve this parabolic P.D.E. by using both explicit and Crank-Nicolson implicit finite-difference methods. Then, we look for algorithms designed for implementing these computations in parallel. Finally, ye evaluate all the algorithms by comparing the numerical results with respect to accu- . racy and wall-clock time of code executions. Codes are executed on Advanced System for Multi-Computer Applications (ASMA) Linux PC cluster. ASMA is located in the Department of Computer Engineering in Bogazisi University, Turkey.