Curve fitting

[Streinerr]

Hello everybody,

Does anyone have suggestions on how to preform a curve fitting using a  double exponent decay, i.e. y=yo+A1exp(-x/t1)+A2exp(-x/t2) ?

Currently I'm using the 7.5 LabView prog.

Thanks,

Y.

P.S I've added a DAT file with the raw data so you'll get a hunch.

[altenbach]

There is no LabVIEW 7.5. Can you chek the correct version you actually have?

Prior to LabVIEW 8.0, the nonlinear fitting tools were not very good. Can you possibly upgrade?

[johnsold]

Actually there was a version 7.5.  It was not released as such.  It was the basis for the LEGO NXT software.  Greg McC talked about it at NIWeek several years ago, I think.

Lynn

[altenbach]

Yes, 7.5 shouldn't be "out there" at all. Let's just hope he actually has 8.5, for example. 😉

[Streinerr]

Dear knigt of the NI kingdom,

Let's say that I've upgraded.

Will the solution resolve from the non-linear sub Vi or there are better tools to handle the fitting job?

I depend on your answer since I do have to convince a prof. to buy the upgrade

Thanks for the quick reply!

S.

[altenbach · Accepted solution]

I've done multi-exponential decays many times and it works well.

Of course you need reasonable starting values for the parameters and there is of some "ambiguity". For example if you swap A1 with A2 and T1 with T2, you'll get an equally good fit.

(If you are at a university, the upgrade will not bee too expensive. Also check if your campus has a campus license.)

[Streinerr]

Thanks for the tip.

Ciao.