Abstract:
In Monte Carlo simulation of weak approximation
of stochastic differential equations, multi-point distributed random
variables play an important role. However, they need highly
efficient implementations to meet the "real-time" requirements
of applications such as the pricing of complex derivative securities.
In this paper a fast and flexible dedicated hardware
generator of multi-point distributed random variables on a Field
Programmable Gate Array (FPGA) is presented. A comparative
performance analysis demonstrates that the hardware solution is
bottleneck-free, retains the flexibility of a traditional software
implementation and significantly increases the computational
efficiency of the overall simulation.
