forum.alglib.net http://forum.alglib.net/ 

real part of fft http://forum.alglib.net/viewtopic.php?f=2&t=414 
Page 1 of 1 
Author:  sambafr [ Wed Aug 17, 2011 11:05 am ] 
Post subject:  real part of fft 
Hello, I am new to Alglib, and i am trying to do an fft of the joint positions recieved via Kinect sensor. For each joint i am trying to make 3 FFT of 1 dimension. ( I know it would be better 1 of 3, but i don't see the point for now, so i rather KISS) I was searching for an fft c# library which would simple enough for me to use, Alglib seems very simple Unfortunately, i can not manage to find a simple way to convert the string fft out in an array of only the real part. I know it seems stupid but because of the +/ and i's, i can only find buggy ways. I get to the array of complex numbers in string format, and i then mess everything. Thank you for your help Samba PS: Rather than buying the commercial part, are small donations via paypal possible? 
Author:  Sergey.Bochkanov [ Wed Aug 17, 2011 8:12 pm ] 
Post subject:  Re: real part of fft 
If you want to get real part of the FFT, you should read output_array[i].x  it contains real part. Is it what you need? P.S. As for donations  it is impossible to accept PayPal payments from Russia, unfortunately. If you want, you can email me and tell specific sum you want to donate, and I will send you a link to the special webform used for nonstandard payments. (I need to know payments sum and currency beforehand in order to prepare this form) 
Author:  sambafr [ Sat Aug 20, 2011 3:26 pm ] 
Post subject:  Re: real part of fft 
Sergey.Bochkanov wrote: If you want to get real part of the FFT, you should read output_array[i].x  it contains real part. Is it what you need? P.S. As for donations  it is impossible to accept PayPal payments from Russia, unfortunately. If you want, you can email me and tell specific sum you want to donate, and I will send you a link to the special webform used for nonstandard payments. (I need to know payments sum and currency beforehand in order to prepare this form) Thank you very much for your answer it was indeed very simple. I took my time because i wanted to be sure that i undestood before answering . If i draw the "power spectrum" of the joint movement via the kinect sensor, therefore i can do: n = total number of recordings ; alglib.fftr1d(timeref, out freqref); Powerspectrum = 2 * Math.Pow((Math.Sqrt(Math.Pow(freqref[i].x, 2) + Math.Pow(freqref[i].y, 2)) / n), 2) Each frequency step = (n / total time) If this is correct, are there any precautions that i should take with n? (should n be a power of 2 for example?) 2:)Concerning donation, it's not much just a way to say thank you, may i pm you for that? 3°) The final objective of my personal project (opensource) is to look for the stability of jointmotion, therefore i would like to use the Lyapunov Exponent to quantify stabilty. I didn' t find any c# library that could help me code that, do you have an advice? EDITT: i saw that alglib can help for nearest neigboor and delay choosing, but i lack the method to go from that to LLE. thanx in advance 
Page 1 of 1  All times are UTC 
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ 