So I am sitting in a small Italian Village, as Oddball said in "Kelly's Heroes" soaking up the rays man.
The rays yesterday were 42 and AC here is an American Dream.
So the students I work with are pushing to some interesting areas in signal analysis for structural vibrations. Unlike most people who publish made up signals, white noise Gaussian -- 2 Hz and 200 seconds with flat Spectral Density, we deal with reality , 1/f squared signals, noisy and Brown noise base that are 800000 points long and contaminated with everyone who walks or drives a bus or a small truck.
Unfortunately the "best" code to use is the Intel IPP for what I want to do , but I am not going into C++, so I will make do with FORTRAN and MKL.
So I was playing with the correlation function and the data supplied by Intel. They give an X sample that is 8 long and y that is 10 offset.
DATA x/1,2,3,4,5,6,7,8/ DATA y/11,12,13,14,15,16,17,18/ DATA y1/1,2,3,4,5,6,7,8/ DATA z/0,0,0,0,0,0,0,0,0,0,0,0,0,0/ DATA e/88,173,254,330,400,463,518,564,448,343,250,170,104,53/ DATA xshape/k/, yshape/k/, zshape/k2/
So if I run the sample the output is e. But if I run the sample x and y1 the output is functionally related to Y through a polynomial.