Clean Numerical Simulation (CNS) for computing reliable long-term predictionof chaos dynamic system is originally proposed by Professor Shijun Liao in2009. Themain idea of CNS is to use high order Taylor series and high precision data to avoidthe numerical noises caused by rounding and truncation.We developed a multiple-precision parallel computation method, based on which,we can use the CNS with high computational efciency and then we can compute com-plex problems in other felds. Based on the multiple-precision parrallel CNS, we com-pute fve chaotic problems derived from Saltzman dynamic system and got very goodresults, thenweaddarandomdisturbationtotheinitialconditionsandstudiedthe2000samples statistically. we found that, random noise of the initial condition can turn outto be a macroscopic randomness. |