berkeleygw / bgwpy Goto Github PK

Currently, when installing BGWpy with HDF5 support, it will set up the appropriate symbolic links for BerkeleyGW calculations using HDF5 filenames, e.g. epsmat.h5 instead of epsmat.

However, BGWpy will set up the symbolic links for Quantum ESPRESSO calculations using non-HDF5 filenames, i.e. charge-density.dat instead of charge-density.hdf5.

This behaviour is unexpected, and BGWpy should use HDF5 for Quantum ESPRESSO as well as BerkeleyGW when it is installing with HDF5 support.

Currently, BGWpy has a config/user_configuration.py file which must be created by the user during package installation to set values for certain runtime variables related to the user's local environment. While some of the variables in this file can be overriden at runtime, e.g. default_mpi, others cannot be overridden through the public API, e.g. use_hdf5.

We have received requests to start distributing BGWpy through package managers (conda, pip); however, the requirement of a user-created config/user_configuration.py file means that BGWpy can not be installed in the automated fashion necessary for a package manager.

One way to circumvent this problem is to allow the user to override all variables set in config/user_configuration.py at runtime through the public API. This would allow us to have a default configuration file that could be used during automated install, which the user could override at runtime to correspond to their local environment.

Hi,

The requirements says: python 2.7 required (Python 3 not supported at the moment).

But as we all know, python 2.x has reached its deadline and retried now. So why not upgrade to the python 3 support?

Regards,
HY

The example scripts for BGWpy are currently bash scripts. While I'm perfectly comfortable with terminal scripts, I'm an old man, and the kids love their Jupyter Notebooks.

A Jupyter Notebook for BGWpy should be added to the documentation, which can be run from start-to-finish for a reasonably generic system (no fancy qsubs or apruns).

A preliminary version of this exists as a GitHub Page at https://wphuhn.github.io/site/index.html, but it needs some spicing up. Some plotting with matplotlib + seaborn, perhaps an example of using scikit-learn to fit an ML model since the kids also love overfitting models.

Hello,

After doing all GW calculations using BGWpy scripts smoothly, I want to use Wannier90 to plot band structures and DOS.

I found a difficulty to define the variables used with Wannier90, such as: nwann, kbounds, klabels, projections,.....

Could you please add a python script showing how to use Wannier90?

Thanks,

Ragab.