| T O P I C R E V I E W |
| altomare |
Posted - 01/03/2005 : 7:26:30 PM
Hi all,
I am trying to fit a function similar to sqrt(Tc-T): since it is not, mathematically, defined for T>Tc, Origin only fits my data for T<Tc and returns -- for T>Tc. Unfortunately if I try to minimize the chi^2 those -- are conted as zeros therefore changing the chi^2 and causing to obtain a very bad fit.
Since the mathematical expression is an approximation and I don't like to restrict the range over which I perform the fit, I was wondering if it is possible to have a fuction defined as follows:
f(x)=experimental data for T>Tc
f(x)= fitting function for T<Tc
Any idea how to implement this?
Thanks a lot
Fabio
|
| 6 L A T E S T R E P L I E S (Newest First) |
| altomare |
Posted - 01/08/2005 : 10:18:08 AM
Mike,
sorry for the late reply. The reason I don't like the data selector tool is that,in my experience, is not very reliable when you need to change the data range or change graphon which to per form the fit.
I haven't had time to try anything else but I will post some info when I will solve this problem.
Thanks
Fabio |
| Mike Buess |
Posted - 01/04/2005 : 3:55:22 PM
If you are fitting from a graph use the data selector tool (Tools toolbar) to define the range before you start the fitter. Then start the fitter and select your dataset. The fitting range will automatically be set to that chosen with the data selector.
Mike Buess
Origin WebRing Member |
| altomare |
Posted - 01/04/2005 : 3:19:50 PM
Mike,
I don't like restricting the fit becasue, when I do it graphically, sometime it is a pain to modify it to include a different range.
I am going to try to define a temproary dataset in a wks and compare that with the data.
If you think of something simpler, please let me know.
Thanks a lot
Fabio
PS I have tried my previous idea but for some reason it doesn't work, I have to think more about it |
| Mike Buess |
Posted - 01/04/2005 : 1:30:28 PM
Fabio,
I assume you are using the Advanced Fitting (NLSF) tool so you can restrict the fitting range appropriately by ignoring the data at T>Tc. Assume, for example, that your X column is equal to Tc at row 101. In the Data Selection section of the NLSF tool highlight the Exp_data dataset and enter 1 and 100 in the boxes on either side of "<= Row <=". Then only the data values for T<Tc will be fitted and you needn't worry about what function to use for T>=Tc.
...Oh, I just noticed in your original post that you don't want to restrict the range. Why not? Seems more appropriate than fitting a dataset against itself.
Mike Buess
Origin WebRing Member
Edited by - Mike Buess on 01/04/2005 1:41:36 PM |
| altomare |
Posted - 01/04/2005 : 12:02:12 PM
Mike
thank you for your reply. You address part of the question but how do I specify the original data in the fitting function box?
Now that I write this I think I can just do
if (t>Tc) y=%(Exp_data,2)
where my experimental data is the second columns of Exp_data.
Am I right?
I am going to try this right away.
Thank you very much
Fabio |
| Mike |
Posted - 01/04/2005 : 11:34:27 AM
Hi Fabio,
Have a look at this forum post from Easwar: http://www.originlab.com/forum/topic.asp?TOPIC_ID=2735
Mike
OriginLab |
