The ABIWAN file (wannier90)

The ABIWAN file (wannier90)#

This notebook shows how to use AbiPy to analyze the output files produced by wannier90 and how to use the ABIWAN.nc file produced by Abinit to interpolate band energies.

As usual, one can use:

abiopen.py FILE_ABIWAN.nc

with the --expose or the --print option for a command line interface and --notebook to generate a jupyter notebook.

Note: The code in this notebook requires abinit >= 8.9 and abipy >= 0.6

Table of Contents#

Let’s start by importing the basic modules needed for this tutorial.

import os

import warnings
warnings.filterwarnings("ignore")  # Ignore warnings

from abipy import abilab
abilab.enable_notebook() # This line tells AbiPy we are running inside a notebook
import abipy.data as abidata

# This line configures matplotlib to show figures embedded in the notebook.
# Replace `inline` with `notebook` in classic notebook
%matplotlib inline

# Option available in jupyterlab. See https://github.com/matplotlib/jupyter-matplotlib
#%matplotlib widget

How to analyze the WOUT file#

Use abiopen to open a wout file (the main output file produced by wannier90):

filepath = os.path.join(abidata.dirpath, "refs", "wannier90", "example01_gaas.wout")

wout = abilab.abiopen(filepath)
print(wout)

---------------------------------------------------------------------------
FileNotFoundError                         Traceback (most recent call last)
Cell In[2], line 3
      1 filepath = os.path.join(abidata.dirpath, "refs", "wannier90", "example01_gaas.wout")
----> 3 wout = abilab.abiopen(filepath)
      4 print(wout)

File /usr/share/miniconda/envs/abipy/lib/python3.12/site-packages/abipy/abilab.py:332, in abiopen(filepath)
    329     return phonopy.load(filepath)
    331 cls = abifile_subclass_from_filename(filepath)
--> 332 return cls.from_file(filepath)

File /usr/share/miniconda/envs/abipy/lib/python3.12/site-packages/abipy/core/mixins.py:55, in BaseFile.from_file(cls, filepath)
     53 """Initialize the object from a string."""
     54 if isinstance(filepath, cls): return filepath
---> 55 return cls(filepath)

File /usr/share/miniconda/envs/abipy/lib/python3.12/site-packages/abipy/wannier90/wout.py:32, in WoutFile.__init__(self, filepath)
     31 def __init__(self, filepath):
---> 32     super().__init__(filepath)
     33     self.warnings = []
     34     self.use_disentangle = False

File /usr/share/miniconda/envs/abipy/lib/python3.12/site-packages/abipy/core/mixins.py:43, in BaseFile.__init__(self, filepath)
     40 self._filepath = os.path.abspath(filepath)
     42 # Save stat values
---> 43 stat = os.stat(filepath)
     44 self._last_atime = stat.st_atime
     45 self._last_mtime = stat.st_mtime

FileNotFoundError: [Errno 2] No such file or directory: '/usr/share/miniconda/envs/abipy/lib/python3.12/site-packages/abipy/data/refs/wannier90/example01_gaas.wout'

To plot the convergence of the wannier cycle use:

wout.plot();

To plot the evolution of the Wannier centers and spread, use:

wout.plot_centers_spread();

Using ABIWAN.nc to interpolate band energies#

ABIWAN.nc is a netcdf file produced by Abinit after having called wannier90 in library mode. The file contains the unitary transformation and other important parameters associated to the calculations. This file can be read by AbiPy and can be used to interpolate band energies with the wannier method.

As usual, use abiopen to open the file:

filepath = os.path.join(abidata.dirpath, "refs", "wannier90", "tutoplugs_tw90_4", "tw90_4o_DS3_ABIWAN.nc")
abiwan = abilab.abiopen(filepath)
print(abiwan)

To plot the matrix elements of the KS Hamiltonian in real space in the Wannier Gauge, use:

abiwan.hwan.plot(title="Matrix elements in real space");

To interpolate the KS energies along a high-symmetry k-path and construct a new ElectronBands object, use:

ebands_kpath = abiwan.interpolate_ebands()

ebands_kpath.plot(title="Wannier-interpolated");

If you need an IBZ sampling instead of a k-path, for instance a 36x36x36 k-mesh, use:

ebands_kmesh = abiwan.interpolate_ebands(ngkpt=(36, 36, 36))

As we are dealing with AbiPy objects, we can easily reuse the AbiPy API to plot bands + DOS:

ebands_kpath.plot_with_edos(ebands_kmesh.get_edos(), title="Wannier-interpolated bands and DOS");

We can also compare an ab-initio band structure with the Wannier-interpolated results. This is useful to understand if our wannier functions are well localized and if the k-mesh used with wannier90 is dense enough.

In this case, it is just a matter of passing the path to the netcdf file containing the ab-initio band structure to the get_plotter_from_ebands method of abiwan. The function interpolates the band energies using the k-path found in the netcdf file and returns a plotter object:

import abipy.data as abidata
gsr_path = abidata.ref_file("si_nscf_GSR.nc")

plotter = abiwan.get_plotter_from_ebands(gsr_path)

Then we call combiplot to plot the two band structures on the same figure:

plotter.combiplot();

As we can see, the interpolated band structures is not completely on top of the ab-initio results. To improve the agreement we should try to reduced the spread and/or increase the density of the k-mesh used in the wannierization procedure.