GlassPy 0.5.3

What's Changed

• Fixed the single-task neural network models. The previous models were not the final models as reported in the publication.
• Fixed conversion of resistivity data when loading SciGlass data (this also fixes GlassNet's resistivity prediction).
• Fixed ViscNet predictions not being deterministic.
• Prediction model files are now downloaded the first time GlassPy is run.
• General code improvements.

GlassPy 0.4.6

What's Changed

• Minor fixes by @drcassar in #14

Full Changelog: v0.4.5...v0.4.6

GlassPy 0.4.5

Bug fixes.

GlassPy 0.4.4

SciGlass data is no longer shipped with GlassPy. The module will automatically download the necessary files when they are needed. Different GlassPy installations (e.g. different virtual environments) now share the same SciGlass files.

GlassPy 0.4.3

Fixes #10.

GlassPy 0.4.2

Minor fix to the documentation.

GlassPy 0.4.1

This version has some bug fixes and changes to the GlassNet class.

GlassPy 0.4.0

Overview:

• LICENSE file now have the lincese for all the data used in GlassPy.
• Now we have a documentation. Check it out at https://glasspy.readthedocs.io/
• glasspy.chemistry has new CompositionType, tools to work with composition arrays, and a featurizer.
• glasspy.data now has the SciGlass class to load data from SciGlass.
• GlassPy now uses the original SciGlass data without any modifications. All the processing is done within GlassPy.
• glasspy.nucleation, glasspy.thermodynamics, and glasspy.relaxation were removed for now. Please use version 0.3 if you want these functionalities. These subpackages will probably come back in the future, but right now I'm focusing on the data and prediction of glass properties.
• glasspy.predict now has GlassNet and ViscNet as available models. GlassPy is now the oficial home of ViscNet. The ViscNet repository will soon be archived.
• new support submodule with support functions.
• glasspy.viscosity was changed and some functionality was removed. It will be back in the future.

GlassPy 0.3

Version 0.3 Release Notes

Initial support for loading SciGlass data. No additional downloads are necessary, the data is provided with the module.