Changelog

All notable changes to this project will be documented in this file. The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

0.1.4 (2023-02-22)

First inofficial release

  • First version of the package on PyPi

  • All main functionality tested

  • Basic documentation

0.2.0 (2023-04-20)

Second inofficial release

  • All important part of the documentation are present

  • This is a test before the first official release 1.0.0

0.2.1 (2023-04-21)

Third inofficial release

  • Test the final process for the first official release 1.0.0

1.0.0 (2023-04-21)

First official release

  • All important part of the documentation are present

  • All main functionality tested

1.0.1 (2023-06-12)

First patch release

  • New notebook to reproduce the results of the paper

  • Fixing some minor bugs

  • Adding fixed stoichiometries in the G matrix

1.1.0 (2024-04-19)

First major update

In the following we refer to an instance of espm.datasets.EDS_espm as spim.

Conceptual changes : * In ESpM-NMF

Syntax changes : * In spim.build_G(), the keyword argument to separate high and low energy lines in the G matrix is now elements_dict instead of reference_elt. The dictionary allows atomic number or chemical symbol notation. * When calling espm.estimators.SmoothNMF, prefer the use of spim.model instead of spim.G to pass the EDXS modelling. * The metadata of the spim can be set in two ways :

  • with the set functions spim.set_analysis_parameters that replace the existing metadata.

  • with the add functions spim.add_analysis_parameters that do not replace the existing metadata.

Conceptual changes : * The espm.estimators.SmoothNMF object can take G as a keyword argument. The argument accepts either a numpy.ndarray (that can be called with spim.G) or an instance of espm.models.EDXS (that can be called with spim.model). * When calling, for example, spim.build_G(elements_dict = {‘Fe’ : 4.0}) the Fe lines are split between high and low energy lines. Now, in the simplex constraint and in the quantification, the low energy lines are ignored.

New features : * The espm.datasets.EDS_espm.estimate_best_binning can be used to estimate the best binning factor to apply on the data before performing the ESpM-NMF decompostion. Use the output (bb) of this function in spim.rebin(scale = bb ) to apply the binning. * An alternative init for the ESpM-NMF decompostion can be activated by executing spim.custom_init = True. It works only when using G = spim.model in espm.estimators.SmoothNMF. * The spim.print_concentration_report was improved thanks to the prettytable package. Statistical errors on the quantifications are now displayed.

1.1.1 (2024-04-25)

Patch of the 1.1.0

  • Fixing the version of traits, hyperspy 2.0.0 is not compatible with traits 6.0.0 and above.