How to do multiple gaussian profiles summation, peak detection and fwhm analysis?

I have 10 gaussians profiles f1...f10(see the text file)and "f" is sum of all. QUE_1)I wanted to find fwhm and plot fwhm lines for each 10 curves. QUE_2) I have used findpeaks function to detect the peaks of all curves,but problem i am facing now is ,it can detect the peak but peak does not move when i change the mean values of one or more curves. QUE_3) How to calculate the fwhm and plot it, when I see two or more curves as a final output (f) QUE_4) how I can change the x and y axis values to some wavelength(x) and reflectivity(y) values. I tried to change X= -1550:0.1:-1545(to see x axis as a nano meter values)but it changes the curves,is there any way to relate x axis as wavelength values without affecting bell curves? QUE_5) How to calculate the peak shift in "f"(in x and y direction) when mean values are changed?

Answer 1

[Akira Agata]

Regarding QUE_1 and QUE_3, please try the following:

findpeaks(f1,'WidthReference','halfheight','Annotate','extents')

Without setting output variables, findpeaks function generates graph with peaks and some additional information if you set 'Annotate' option to 'extents'.

[Akira Agata]

Sorry for this figure with some Japanese phrases. This is due to my MATLAB's language settings.

In addition, I would be happy if you could explain more details on your other QUEs.

[Tusharkumar Sorathiya]

Thanks a lot Akira for the answer, well till now I have solved QUE_4. And I am presuming I will find solution for QUE_2 & QUE_5 accordingly if I get the results as you answered for QUE_1/3. I came to know the solution you said should work in MATLAB. The problem is, I am using Octave-4.2.1 (GUI) now at my office. And it seems findpeaks function works fine,but it says " argument 'WidthReference' is not a valid parameter" when I try your given solution. 1) I guess for Octave this parameters must have different names! Any suggestions? I am attaching the findpeaks function which I am using currently for your reference. 2) Could you please provide me your findpeaks.m function so I can try it in matlab at my university computer if it works.

[Tusharkumar Sorathiya]

Hello Akira, again thanks a lot. Any idea how i can do the same(which you answered) in OCTAVE?

[Akira Agata]

Hi Tusharkumar-san,

I regret to say that I'm not so familiar with OCTAVE. Looking at the following help page, it seems that OCTAVE's findpeaks function does not have exactly the same option used in my above post. But, using "parabol", "height" and "baseline" values stored in the 3rd output argument, maybe you can calculate approximate FWHM value for each pulse.

https://octave.sourceforge.io/signal/function/findpeaks.html