Differences between NAG and FFTPACK

• FFTPACK expects and returns data in a different format to NAG.
• FFTPACK includes initialisation routines (PDA_RFFTI and PDA_CFFTI) which should be called prior to the other routines, but which don't need to be called again until the size of the data array changes. There are no equivalent NAG initialisation routines. The main NAG FFT routines (e.g. C06FAF, etc.) do this initialisation each time they are called, irrespective of the array size.
• FFTPACK has separate forward and backward transform routines, whereas NAG only has forward routines (backward transforms are performed by using complex conjugation with the forward transform). This means that there is probably no need to supply equivalents to the complex conjugation NAG routines (a trivial operation anyway).
• FFTPACK can accept arrays of any length, whereas NAG puts some restrictions on the array length (no prime factor larger than 19 allowed in the array size, and the total number of prime factors must be less than 21).
• FFTPACK routines require a differently sized work space array.
• FFTPACK has no error checking.