Analysis And Research On The Rapid Realization Of Financial Derivatives Pricing Models Based On Heterogeneous Computing Platforms

This work aims to use the CPU-GPU heterogeneous computer hardware platform to quickly implement European option pricing in the field of financial derivatives with large amounts of data.The calculation function provides a guarantee for the fast and accurate results of the work.In this work,the common Monte Carlo simulation method and the one-step Monte Carlo simulation method are used to price European options.During the entire pricing process,both simulation methods were serially calculated and parallelly calculated.The simulation results are consistent with the large In the case of the law of numbers,European option simulation prices have a convergence effect and high simulation efficiency.The most important part of using the Monte Carlo simulation method is its random number part.In order to work stable and fast at the end of the work,the generation of normal random numbers is crucial.This work uses the multiplicative congruence algorithm plus the Box-Muller algorithm to generate the normal The distributed random numbers are used in two Monte Carlo simulation methods.To use ordinary Monte Carlo simulation methods to price European options,first establish its serial pricing model in the working environment and obtain the results,and then transplant the model to a parallel working environment to achieve Monte Carlo parallel simulation results Through the comparison of serial simulation and parallel simulation,the analysis results show that the pricing results of the single-threaded,multi-threaded and parallel modes are consistent,and the error of the results gradually decreases with the increase of the parallel scale.The one-step Monte Carlo simulation method uses the SABR model to correct the common Monte Carlo simulation method.In European option pricing,it is more accurate than the common Monte Carlo simulation method.Using this method to simulate European option pricing,when using the same data,The calculation results using this simulation method are more accurate.The innovation of this work is to use the one-step Monte Carlo simulation method on heterogeneous computing platforms,and compare the method with serial and parallel simulations to accurately and quickly obtain pricing results.