search

vasole / pymca

PyMca Toolkit git repository
Other
58 stars 54 forks source link

Basic questions about fitting #10

Closed licode closed 10 years ago

licode commented 10 years ago

It will be great if you can answer some basic questions about pymca. I feel sorry if the questions are kind of scattered.

  1. How did you deal with branching ratio? Did you give users full control of those values, or the ratios are fixed, or calculated based on incident energy? Any recommendations?
  2. Do you only use gaussian functions for different lines of elements, or other kinds of functions are implemented? Can you point out to me the place the functions are included?
  3. Do you assume background functions as polynomial to fit, or just use snip method to remove it directly? Any idea which one works better?
  4. Are the peak width fitted globally(function of energy), or locally (every peak width is independent fitted), or both?
  5. Do you also set the peak position as a fitting parameter? I know we need to do energy calibration first.

Thanks!

licode commented 10 years ago

cc @tacaswell @ericdill

vasole commented 10 years ago

Hi,

1 - The ratios are calculated from the complete experimental setup. Beam hardening, matrix effects, attenuation between sample and detector, detector efficiency, .... The starting theoretical parameters are in the files K/L/MShellRates.dat they are in ASCII. The user can easily visualize them within the program in several ways. The simplest is PyMca Main Menu -> Tools -> Elements Info

2 - Hypermet function and (Pseudo-Voigt) function. They are coded in C, but SpecfitFunctions.py calls them.

3 - The SNIP is will not give you any trouble during the fitting. It is the one I use. If you have poor statistics and you can manage with a low order polynomial you will get faster fits.

4 & 5- I think you should read the published paper. If you search internet for "A multiplatform X-ray fluorescence toolkit" you should get the pdf..

The initial calibration is not needed to be very accurate. The program will refine it.

Armando

vasole commented 10 years ago

The official mailing list of PyMca is at sourceforge. I have only moved the repository and issue tracker. It would have been more productive to have posted the questions there because the audience would have been much larger. The majority of PyMca users are not programmers, so few of them have github accounts.

Best regards,

Armando