Dear all, One colleague of mine brought up this to my attention. He tried a simple Fourier transformation C code on different machines (PowerPC based MAC G4, Intel based MAC G5, and PCs). The computations yield different results, sometimes on the third significant place. In his Fourier transformation implementation, a standard complex function is called. On MAC boxes, he used GNU C compiler; on PC, MS visual C compiler was used. I have done a little web survey on this issue, it seems that this was reported by others, too; there have been suggestions that it is much more reliable to handle real or imaginary part separately. Since in the Fourier transformation for EXAFS function, complex functions have to be handled too. One thing I would like to know from our Ifeffit Gurus: Are the real and imaginary parts of EXAFS functions been handled separately? Thanks for your attention! Best, Ying ========================== Dr. Ying Zou Research Associate, Physics/UWM 1900 E. Kenwood Blvd. Milwaukee, WI 53217 ==========================