The following tutorial will guide you through the basic steps of using the TBmodels package: Creating, evaluating and saving a Model instance.

Creating a Model instance

The Model class describes tight-binding models within TBmodels. Instances of this class can be created in various ways, some of which I will describe here.

Using Wannier90 output

First off, a model can be created from the output of a Wannier90 calculation. Note that use_ws_distance = .true. should be used (available since Wannier90 2.1) to get the correct tight-binding model. Wannier90 creates several files that can be read by TBmodels:

  • The *_hr.dat file contains the hopping terms. Use write_hr = .true. to generate this file.

  • The *_wsvec.dat file contains correction terms for the lattice vectors of the hopping terms.

  • The *_centres.xyz file contains the Wannier centers. They can be used to determine the positions of the orbitals in TBmodels.

  • The Wannier90 input file *.win can be used to determine the unit cell dimensions.

To generate a TBmodels Model instance, the *_hr.dat file is required, and the other files can be added optionally. However, it is recommended to at least also use the *_wsvec.dat file.

import tbmodels

# use only *_hr.dat
model = tbmodels.Model.from_wannier_files(hr_file='path_to_directory/wannier90_hr.dat')

# use all files
model = tbmodels.Model.from_wannier_files(

Constructing models manually

Alternatively, a Model instance can be created directly using the constructor. The following example shows how to create a model with two orbitals. The orbitals have on-site energies 1 and -1 (the unit can be arbitrary, but must be consistent), and there is one occupied state. The system is three-dimensional, and the orbitals are located at the origin and [0.5, 0.5, 0.] (in reduced coordinates), respectively.

In a second step, hopping terms between the two orbitals (nearest-neighbour interaction) and between the same orbital in different unit cells (next-nearest-neighbour) are added using the add_hop() method.

# (c) 2015-2018, ETH Zurich, Institut fuer Theoretische Physik
# Author: Dominik Gresch <greschd@gmx.ch>

import tbmodels
import itertools

model = tbmodels.Model(
    on_site=[1, -1], dim=3, occ=1, pos=[[0.0, 0.0, 0.0], [0.5, 0.5, 0.0]]

t1, t2 = (0.1, 0.2)
for phase, R in zip([1, -1j, 1j, -1], itertools.product([0, -1], [0, -1], [0])):
    model.add_hop(t1 * phase, 0, 1, R)

for R in ((r[0], r[1], 0) for r in itertools.permutations([0, 1])):
    model.add_hop(t2, 0, 0, R)
    model.add_hop(-t2, 1, 1, R)

Evaluating the model

Once created, a Model instance can be evaluated at different k-points in the Brillouin zone using the hamilton() and eigenval() methods. These methods take a single k-point, given in reduced coordinates, as argument.

print(model.hamilton(k=[0., 0., 0.]))
print(model.eigenval(k=[0., 0., 0.]))

Saving the model to a file

There are different ways of saving the model to a file. To save the model for later use, I recommend using the to_hdf5_file() method. This will preserve the model exactly as it is.

model2 = tbmodels.Model.from_hdf5_file('model.hdf5') # model2 is an exact copy of model

If compatibility with other codes operating on Wannier90’s *hr.dat format is needed, the to_hr_file() method can be used. However, this preserves only the hopping terms, not the positions of the atoms or shape of the unit cell. Also, the precision of the hopping terms is truncated.

model3 = tbmodels.Model.from_hr_file('model_hr.dat') # model3 might differ from model

Finally, the Model class is also compatible with Python’s built-in pickle module. However, data saved with pickle may not be readable with different versions of TBmodels since pickle serialization depends on the specific names of classes and their attributes. The use of pickle compatibility is to enable multiprocessing with TBmodels.

Command-line interface

TBmodels has a built-in command line interface, which is designed to perform common operations, such as converting Wannier90 output into a TBmodels file. It can be accessed via the tbmodels command, and its documentation can be accessed with the --help flag or in the reference section. For example, tbmodels --help gives a list of possible commands, and tbmodels parse --help gives specific information about the parse command.


The command line commands may change in future releases, because this feature is still in early development.