Fitting with an integral function

[Zizou]

Hello,

I would like to fit my data points with a function defined by an integral of the following kind:

y(x)= INTEGRAL (from 0 to infinity) [f(x,t)*Gauss(t,t0,Sigma)]dt

where Gauss (t,t0,Sigma) is the gaussian distribution with mean value t0 and Sigma is half width broadening while f depend on x and t. The fitting parameters are of course t0 and sigma

Is it possible with labview ? I have no experience fitting with labview ? Does anybody has a piece of VI that could help me doing that ?

Thanks for you help

[altenbach]

You can fit anything you want as long as you have...

1. a list of parameters you want to fit
2. a suitable model that calculates the function based on these parameters for a set of x values
3. data you want to fit for the same set of x values.

Easiest would probably be to use nonlinear curve fit with the VI model. What version of LabVIEW do you have?

[Zizou]

1. a list of parameters you want to fit
2. a suitable model that calculates the function based on these parameters for a set of x values
3. data you want to fit for the same set of x values.

Easiest would probably be to use nonlinear curve fit with the VI model. What version of LabVIEW do you have?

I have labview 7.0 Express. Do you mean using the Levenberg Marquardt non linear curve fit ? I suppose that the integral as a function of x, t0 and sigma can be evaluated by labview and then trying to fit tthe data points with such a new created function ?

thanks

[altenbach]

"Nonlinear curve fit" is new for LabVIEW 8.0 and is vastly improved over the old levenberg marquardt routines.

There are also some easier to use levenberg-marquardt routines that I modified long ago and one example in LabVIEW 7.0 can be found here:

http://forums.ni.com/ni/board/message?board.id=170&message.id=129944#M129944

Maybe you can re-use some of it. 😄

[Zizou]

I don't  have labview .

I still have problem to understand whether or not I could use labview for fitting my data points. If I look at a Levenber-Marqurdt.vi, the model has to specified as a string. The function I want to use for fitting is defines as an integral, F(x,a)=integral (-infinity, +infinity)G(t,a) dt where I want to determine a by fitting.

This expression cannot be written as a string !

How should I implement such a function ? Possible or not

Thanks

[altenbach]

---

@Zizou wrote:
I don't  have labview

---

Well, you definitely need labview 🙂

Look for Nonlinear Lev-Mar fit. I uses a subVI model. My modified VIs are based on it, but implement a few important changes so it can be used for more general cases.

The link I gave you has LabVIEW 7.0 versions of my VIs. You said you have LabVIEW 7.0. Can you open them?  (There is even a LabVIEW 6.1 version!)

[Zizou]

Sorry, actually I have Labview 7.0.

Ok, I will try with Nonlinea Lev-Mar-fit

Thanks

[altenbach]

---

Zizou wrote: Sorry, actually I have Labview 7.0.

---

So, the example at my link above should work for you.