Re: Fitting with leasqr and uncertainties

[CdeMills]

st.michal88 wrote
> Hello everybody,
> 
>  i am pretty new with Octave and I need some advice. I am using leasqr
> function to fit my experimental data with exponential function. The
> problem is that uncertainties of parameters are very important to me. I
> diged up internet with this question and didn't find the answer. What I
> want to get at the end is a set of parameters in form: parameter +-
> uncertainty. Is it possible with this function? I found out also about
> nonlin_curvefit but it doesn't give me the answer eighter. Thank you in
> advance for your help.

Hello,

fitting exponentials to real signal is very difficult for a number of
reasons; one of them is the noise sensitivity. I refined one such approach
over the years; it uses the Prony method. Basically it expects some vector
to either contains a linear combination of exponentials / sines / damped
sines; either be the response of a LTI system to some input made of steps
whose width is greater (in terms of number of samples) than the number of
basic terms (exponentials / sine /...)  The function also estimates
covariance matrices over the non-linear parameters (the time constant /
frequency terms) and the linear ones (the associated weights). As already
discussed, the on-diagonal terms are equivalent to standard deviation
estimates.

I was thinking since a few time to make a toolbox of this function. Could
you describe your experimental signal ? 

Regards

Pascal



--
View this message in context: 
http://octave.1599824.n4.nabble.com/Fitting-with-leasqr-and-uncertainties-tp4661189p4661202.html
Sent from the Octave - General mailing list archive at Nabble.com.