| Author |
 Topic  |
|
|
helporigin8
29 Posts |
 Posted - 12/16/2016 : 09:46:59 AM
|
Hello,
Actually, I would like to do following:
I have tree columns with data that should simulated certain process (x1, x2 and x3) and I would like to make a function y=a*x1+b*x2+c*x3 which would take data from these tree columns and sum them in different proportions in order to make best fit to my experimental data.
I am not sure is it possible to do with OriginPro 8.5 (version that I have), but I searched for solution on internet and did not find solution.
Thank you,
A. |
Edited by - helporigin8 on 12/16/2016 09:48:56 AM |
|
|
JacquelineHe
287 Posts |
|
|
helporigin8
29 Posts |
Posted - 12/19/2016 : 05:01:07 AM
|
Yes, it is what I am looking for. But now I have another question:
If my function with multiple independent variables is P=a*x1+b*x2+c*x3, how could I set condition a+b+c=1? |
Edited by - helporigin8 on 12/19/2016 05:02:01 AM |
 |
|
|
snowli
USA
1386 Posts |
|
|
helporigin8
29 Posts |
Posted - 12/19/2016 : 7:25:32 PM
|
Sorry,
But still, it seems that it is not what I need. As you can see from attached picture, I want to fit my experimental data from columns E, F and G with function that is defined by columns A, B, C and D, i.e.: P=a*col(B)+b*col(C)+c*col(D), where a+b+c=1 and a,b,c>0.
 |
|
|
|
helporigin8
29 Posts |
Posted - 12/19/2016 : 7:29:21 PM
|
| In way that you suggested me, if I am not wrong, I can only to calculate new function independently from experimental set of data, but I want to calculate this function in that way to observe best fit to exp. data. |
|
|
|
Sam Fang
292 Posts |
Posted - 12/21/2016 : 12:58:55 AM
|
I didn't understand your question about "fit my experimental data from columns E, F and G with function that is defined by columns A, B, C and D". Can you explain it more detailed? e.g. You can show the formula for columns E, F and G.
Sam
OriginLab Technical Services |
|
|
|
helporigin8
29 Posts |
Posted - 12/21/2016 : 04:48:46 AM
|
| Yes, I have to fit experimental data with this function: P=a*col(B)+b*col(C)+c*col(D), column A is x-axis for all three y-axis (columns B, C and D) and a, b and c should be free parameters of the fit. |
|
|
|
helporigin8
29 Posts |
Posted - 12/21/2016 : 04:52:28 AM
|
| And these free parameters I have to deduce from the best fit to the experimental data. |
|
|
|
YimingChen
1606 Posts |
Posted - 12/21/2016 : 3:19:11 PM
|
If you want to apply linear constraints to your fitting parameters, e.g., A+B+C=1, open up the non-linear fitting dialog (ctrl+y). Under code tab, inside constraints branch, type in "A+B+C=1". Enable "enable linear constraints" checkbox. You can refer to this tutorial page for more detail:
http://www.originlab.com/doc/Tutorials/Fitting-FixThroughPoint
|
|
|
|
Hideo Fujii
USA
1582 Posts |
Posted - 12/21/2016 : 4:23:31 PM
|
Hi,
If a+b+c=1, then your fitting function can be rewritten as:
P=a*col(B)+b*col(C)+(1-a-b)*col(D)
thus, no need of that constraint, I think, but only a,b>0 and (a+b)<1 .
--Hideo Fujii
OriginLab |
Edited by - Hideo Fujii on 12/21/2016 4:28:11 PM |
|
|
|
Sam Fang
292 Posts |
Posted - 12/21/2016 : 9:36:34 PM
|
Do you mean P in the function is column F, x (A) for B, C and D is different from x (E) of F and you also want to fit with Y error G?
The snapshot you attached is just part of data? I wonder the last value of column A. Is it greater than 4.41 in column E? If so, maybe you can interpolate first.
Sam
OriginLab Technical Services |
|
|
| |
Topic |
|
Fitting data with function parameterized in coulom