摘要
本文研究了一类素因子分解ＦＦＴ算法的指数映射，用这些指数映射方法可以构成一类新的素因子分解ＦＦＴ算法（ＰＦＡ）。此类算法可以用非同址的方式实现，也可以用同址的方式实现；既可以输入输出皆为同一顺序而不需要混序，也可以输入输出不为同一顺序而需要混序。本类算法具有新的算法结构，在计算每一维的小点数ＤＦＴ时，需要变换数据模块的地址。理论分析与计算机仿真实验表明，新算法不仅保持了传统ＰＦＡ［３］能够节省存贮量、提高运算速度的优点，而且具有可以顺序输入输出、同址计算、无需混序操作的特点。
This paper presents a new prime factor algorithm of FFT in which data can be calculated in place or non in place and in the order with no input-output scrambling or in the order with scrambling. The algorithm presented is different in structure from the traditional prime factor algorithm and requires address translation of intermediate data module while each one dimensional small scale DFT of multidimensional DFT is calculated. Analysis and experiment on the computer show that, as compared to the traditional prime factor algorithm, scrambling operation is not necessary in the new algorithm, in which computation can be mode in place, in sequential order, with less storing units and improved computing speed.
出处
《声学学报》
EI
CSCD
北大核心
1995年第6期401-407,共7页
Acta Acustica
基金
国家自然科学基金
