belaloui / thin-film-mlines-fit Goto Github PK

Get rid of the multiple hardcoded fitting functions and write a single one with dynamic parameters where the fixed parameters are specified with their respective values, and the free parameters are specified with their boundaries.

It may be useful to load these parameters from a file.

Save the config file with the results.

• Put the fitting functions in a separate file.
• Add a selection for the fitting methods: between Scipy and PyGad.
• Add a selection for Rp and Rs.
• Add a selection for background removal. (will need a transfer curve)
• Add a verbose mode. (With callback functions in the fitting part)
• Print the results at the end of the script.
• Save the data in a file.
• Fix the plots accordingly.

Parameter names, used files, etc.

Self-explanatory.

A separate script must handle the drawing of raw data, truncation of the curves, and the drawing of the pre-processed data.

The model's file should include the fixed parameters and the bounds for the free parameters.

Explain in some detail what the problem we're solving is, and what we have implemented in the code, as well as a small tutorial on how to use it.

Probably Apache 2.0.

line 126.

To generate the fitted curve, we should average over the parameters, not the curves, i.e, we take the mean parameters as the fitted curve's parameter.

This should include the curve data, polarization, fitting method, curve correction with background data, and so on.

An explicit list of tasks needs to be written down here.

Use one of the metrics that are used in the literature.

The newest data needs to be added to the repository both for availability and to make testing the code as it is refactored easier.

Code from SO:

popt, pcov = curve_fit(f, xdata, ydata)
residuals = ydata- f(xdata, *popt)
ss_res = numpy.sum(residuals**2)
ss_tot = numpy.sum((ydata-numpy.mean(ydata))**2)  
r_squared = 1 - (ss_res / ss_tot)