| Author |
 Topic  |
|
|
menhart@ii
USA
23 Posts |
 Posted - 02/24/2017 : 1:44:13 PM
|
is is possible to use the multiple peaks NLLS function on multiple data sets?
i want to globally link xc over several data sets.
the multiple peak fitter works well on single data sets. but when i move to multiple data sets so i can link the, it kicks me back down to a single gaussian. do i need to program in my own multiple gaussian function? that seems silly... |
|
|
yuki_wu
896 Posts |
|
|
menhart@ii
USA
23 Posts |
Posted - 03/04/2017 : 3:17:50 PM
|
no not really.... i have been using origin since v1 (i.e. when it was still part of microcal!) and i know about global fitting. this KB entry describes that,a nd how to use script to do to that.
that is fine. i understand i can code my own multiple gaussian function, independent of the Origin provided one, and do that. but i assume the built in one is perhaps faster and more efficient, and id rather no re-invent the wheel if not necessary.
so my question is more about how multiple peaks is implemented in O2016 (i will soon upgrade to O2017 if that makes a difference). when i fit a single data set to multiple gaussians using the built in multiple peaks instance of the guassian function, it works well. but as soon as i load multiple data sets, it automatically revert to single guassian, and i cannot do multiple guassian in a global manner. i would like to link xc and perhaps w across all data sets
so - do i have to code my own user function? |
 |
|
|
yuki_wu
896 Posts |
Posted - 03/06/2017 : 03:42:52 AM
|
Hi,
I am sorry that we cannot do sequential fit with a multiple peaks function at present, but this feature is targeted to be included in a future release. It may be achieved by an advanced APP, so keep an eye out for this app's release.
However, there is a workaround. For example, I assume that every dataset has 3 Gaussian peaks:
1. New a user defined fitting function
2. Set the variables and parameters
Independent Variables: x
Dependent Variables: y
Parameter Names: y0,xc1,w1,A1,xc2,w2,A2,xc3,w3,A3
Function Form: Equations
3. Define the fitting function in Function box by quoting built-in Gauss functions:
y = y0 + nlf_Gauss(x, 0, xc1, w1, A1) + nlf_Gauss(x, 0, xc2, w2, A1) + nlf_Gauss(x,0, xc3, w3, A3)
4. Save the function
As a result, you can perform global fitting and then share the parameters in multiple datasets by using this function. Note that properly initialized parameter is the key to success.
If this workaround dosen’t help, you could send your data to us (tech@originlab.com), so that we can try more on your data.
Regards,
Yuki
OriginLab
|
|
|
|
menhart@ii
USA
23 Posts |
Posted - 03/08/2017 : 2:22:29 PM
|
thx yes - i have done that.
or almost that - i wrote my own N*Guass functions. maybe the built in one is faster, but either way - speed is not the issue.
i just had to implement {2Gauss, 3Gauss 4Gauss, 5Guass...} functions so it's not terrible convenient. |
|
|
| |
Topic |
|
|
|
multipel gaussian/ multipel data sets