Author |
Topic  |
|
oszkoa
Hungary
47 Posts |
Posted - 03/19/2004 : 4:11:24 PM
|
Hi, I'd like to fit my measurement data with more than one functions of the same type (e.g. Gaussians). Unfortunately I could not insert an image into the forum, so I try to explain the situation: my spectrum is usually the sum of several overlapping peaks (Gaussians) and what I try to do is to resolve this measured curve into individual peaks. These peaks correspond to different states of the sample that simultaneously exist. Saying it in other words: what i got is the envelope(or wrap - as you like it) of several Gaussians and I try to find the constituents. I should also be able to apply some constraints , like the position, the half-width and intensity ratio. The Analysis/Fit Multi-peaks/Gaussian does something similar but it is not possible to apply constraints. Does Origin have the ability to do that?
Oszkoa |
|
easwar
USA
1965 Posts |
Posted - 03/19/2004 : 4:40:09 PM
|
Hi Oszkoa,
Use the Advacned NLSF tool. Then, when you select a peak function such as gaussian, the menu item Options->Replicas will become available. Use that menu item to bring up the replicas page. Then you can set the number of replicas (how many gausssians you want to fit the data with, for example). On the right side window, the paramters will be listed. You can double-click to make any of them shared between the peaks (for example, fit with many gaussians of same width).
Then, you can place further constraints on the parameters using the NLSF dialog under the Options->Constraints menu.
The Fit Multiple Gaussian menu item uses this, but that is for a quick fit of multiple gaussians that does not allow constraints and sharing, but all can be done using the Adv. NLSF tool.
Easwar OriginLab When you select a peak function such as Gaussian in the [/quote]
|
 |
|
oszkoa
Hungary
47 Posts |
Posted - 03/21/2004 : 3:41:12 PM
|
Hi Easwar,
Thanks for your help. Now it looks like working, though it was a bitter struggle. May I have some further questions: Is it possible to save the parameters of a fit in such a way that they can be used as the initial parameters of another fit? It would be useful when fitting a series of measurement data (output of 1st fit is the input of 2nd fit; output of 2nd fit is the input of 3rd fit and so on). Is it possible to apply other than linear background (I think of Shirley background)? Does Origin's PFM gives much more freedom in setting fitting parameters than the Advanced Fitting Tool?
Oszkoa |
 |
|
easwar
USA
1965 Posts |
Posted - 03/22/2004 : 11:06:35 AM
|
Hi Oszkoa,
The PFM does have better capabilities in terms of baseline subtraction, peak finding, and storing and re-using fit information etc. You can download a demo version of the PFM and give it a try with your data. http://www.originlab.com/index.aspx?s=12&lm=+63
Note that PFM is available currently for free with OriginPro.
Also, you could write script or Origin C code for the NLSF to perform customized fitting that meets your needs.
Easwar OriginLab
|
 |
|
oszkoa
Hungary
47 Posts |
Posted - 03/22/2004 : 2:58:57 PM
|
Hi Easwar,
I downloaded not only the demo but also the PFM manual. My first impression is that fitting process with PFM seems easier than with the Advanced Fitting Tool. And more, it is possible use multiple Voigt functions (it was not possible to create replicas from Voigt in NLSF). All in all, I have to talk to my boss :-)
Oszkoa |
 |
|
|
Topic  |
|
|
|