need help with non linear regression

[mayyan]

input: X= [X1 , X2 ,...,Xn] ; Y= [Y1 , Y2 ,...,Yn]

I want to find a solution of the type:
y = C1*exp{(a+ib)t}+C2*exp{(a-ib)t} or y = D1*exp{-at}*Sin(bt+p)

( which is the solution of: y'' + A0*y' + A1*y = 0 with complex roots ). I want to do it using regression to best fit the input.

Does someone know a book ( preferably Ebook) or internet site with an algorithm? Or have other idea how to solve it using different method?

please let me know
thanks
Mayyan

 

[ATLAJ]

try Matlab with Curve Fit Tools

type on workspace: cftool

you can find linear polynomial , quadratic polynomial , cubic polynomial and 4th degree to 9th degree polynomial fits.

I always use this toolbox. It´s very powerfull e easy.

ATLAJ

 

[mayyan]

Thanks for replying but the phisical nature of my problem make it look like:

y = D1*exp{-at}*Sin(bt+p)

and since my sampling points are few and far from each other the polinomial way want give me satisfactory solution.