From: Dunno on
Hi all,
I am trying to get uncertainties of best fit parameters from fminsearch function. It will be easy if I have a normal Matlab script as fit function, but I use Matlab to call a Linux program to calculate signal from input parameters. So, I created a Matlab function with some input parameters, the output is calculated using the linux program based on the input parameters.
The fit is converged but I am not sure how to get a reliable uncertainties of the fit parameters. I have used pair correlation method but it takes very long time and I don't know if it is correct. Any suggestion is appreciated.
Best,
Thai
From: us on
"Dunno" <pmctkr(a)yahoo.com> wrote in message <hpkndf$n6r$1(a)fred.mathworks.com>...
> Hi all,
> I am trying to get uncertainties of best fit parameters from fminsearch function. It will be easy if I have a normal Matlab script as fit function, but I use Matlab to call a Linux program to calculate signal from input parameters. So, I created a Matlab function with some input parameters, the output is calculated using the linux program based on the input parameters.
> The fit is converged but I am not sure how to get a reliable uncertainties of the fit parameters. I have used pair correlation method but it takes very long time and I don't know if it is correct. Any suggestion is appreciated.
> Best,
> Thai

how can any CSSMer be of any help you if you're using an external program(?)...
THINK(!)...

us
From: Dunno on
Well, it is basically a matlab function for fit. Since it does not work with cftool to give fit uncertainty, I use fminsearch and it works. However, I don't know how to get a good uncertainty out of it.
Btw: what is CSSMer?

"us " <us(a)neurol.unizh.ch> wrote in message <hpkoqh$l7i$1(a)fred.mathworks.com>...
> "Dunno" <pmctkr(a)yahoo.com> wrote in message <hpkndf$n6r$1(a)fred.mathworks.com>...
> > Hi all,
> > I am trying to get uncertainties of best fit parameters from fminsearch function. It will be easy if I have a normal Matlab script as fit function, but I use Matlab to call a Linux program to calculate signal from input parameters. So, I created a Matlab function with some input parameters, the output is calculated using the linux program based on the input parameters.
> > The fit is converged but I am not sure how to get a reliable uncertainties of the fit parameters. I have used pair correlation method but it takes very long time and I don't know if it is correct. Any suggestion is appreciated.
> > Best,
> > Thai
>
> how can any CSSMer be of any help you if you're using an external program(?)...
> THINK(!)...
>
> us