Supplementary material: Dynamical mean-field theory of the Anderson–Hubbard model with local and non-local disorder in tensor formulation

Supplementary material
======================

This is the supplementary material to the article
``Dynamical mean-field theory of the Anderson–Hubbard model with local and
non-local disorder in tensor formulation``.

It contains the DMFT data, the scripts to produces the figures, and some numerical
tests for some of the given equations

Scripts
-------

`Python3` (at least 3.6) is required to run the scripts,
the necessary dependencies are specified by `requirements.txt`::

   $ pip install -r requirements.txt

In case there are dependency problems, pipenv can be used to reproduce the
exact environment used to run the scripts::

   $ pip install pipenv
   $ pipenv install
   $ pipenv run shell

Data
----

The directory `DMFT_data` contains the final iteration of the DMFT calculations
shown in the article. The directories are named after the parameters,
where `U` is the Hubbard interaction, `V` the on-site energy, and
`tab` the hopping parameter between components `A` and `B`.
If parameters differ for the components they are given as `Ua`, `Ub` …

Each directory contains the subdirectories `lattice_output` with the (local)
lattice quantifies and `impurity_ouput` with the results of the effective
impurity models (SIAMs).

The data is provided as `HDF5` files, it can be conveniently processed and
visualized using Python's xarray module.
A convenient `open_dataset` method is provided in `plotting_scripts/dataio.py`.

Figures
-------

The directory `plotting_scripts` contains the Python scripts to generate the
figures. They are prefixed with `figN_` according to the figure numbers.

Tests
-----

The directory `test_scripts` contains some numerical test for formulas given in
the article. The files prefixed with `test_` can be executed by running pytest.

The scripts prefixed with `frequency_expansion_` provide only a visual check
of the high-frequency expansion, they are directly run with `python`. You can
verify by eye that the BEB effective medium converges against the high-frequency
moments given in the article.