semicon - k·p simulations made easy

Note: this is work in progress, comments and ideas are more than welcomed!

The goal of this package is to provide easy to use framwerok for performing k·p simulations.

This package is suppose to remove an overhead and a boiler plate that appears when doing the k·p simulations. It is suppose to make things easier, therefore userfriendly interface is a priority.

Models

Models are based on the k·p Kane Hamiltonian, symmetrized following Burt-Foreman approach. Users should be able to choose which components (Zeeman, Dresselhaus, strain, ...) and which bands ('gamma_6c', 'gamma_8v', ...) shall be included.

Parameters

Parameters bank should include most common semiconducting materials. Helper methods should handle parameter renormalization based on included bands and way to avoid spurious solutions.

Helper functions

Goal of this framework is to make k.p simulations easy. However, building the actual Kwant system and peforming the simulation should be done by user and this library only provide helper tools that are specific to problems encountered in k.p simulations

Requirements

• This package is suppose to work with latest stable release of Kwant.
• This package depends on recently released SciPy 1.2 for some of its optional functionality. It will work with earlier release but appropriate warning will be rendered.
• Because project is based on discretizer sympy is required, however, due to compatibility issue it must be in version lower than 1.2. For now on the recommended version is 1.1.1.

Installation (from pip into conda environment)

The easiest way to install semicon is to create fresh conda environment

conda env create --name semicon kwant=1.3.3 sympy=1.1.1

and install semicon using pip

conda activate semicon
pip install semicon

to test your installation do

python -c 'import semicon; semicon.test()'

Optionally, install Scipy 1.2 to benefit from rotation functionality.

pip install scipy==1.2

Building from sources

As this package is pure python standard python installation from sources with

python setup.py build
python setup.py install

should be working without problems. The only non-trivial dependency, Kwant, that could cause problems if not installed beforehand can be either obtained via conda or any other installation means explained on its homepage.

Direct installation from git is also possible (and favoured as long as conda package is not out there):

pip install git+https://gitlab.kwant-project.org/semicon/semicon.git

Note that due to active development master branch may not be stable. To install version that has been already used in research project use

pip install git+https://gitlab.kwant-project.org/semicon/semicon.git@v0.2.0

Development

We use pre-commit to keep a consistent code style, so pip install pre_commit and run

pre-commit install

to install the hooks. These will then automatically run upon each commit.

Tips about developing inside docker container

One can easily use a rafalskolasinski/science Docker container for a development of this project. Assuming that semicon folder is ~/work/semicon do:

docker pull rafalskolasinski/science:semicon
docker run -d -p 8888:8888 --name semicon \
    -v ~/work/semicon:/src -v ~/work/semicon/notebooks:/home/jovyan/work \
    rafalskolasinski/science:semicon

This will mount source code in /src and project notebooks in ~/work inside the containier. It will also start jupyter notebook server running at localhost:8888. You will need to read jupyter's server token with docker logs semicon to access the server.

You can now use docker exec semicon build and docker exec semicon test to build the package or run the tests respectively.

You can enter bash inside the container by running

docker exec -it semicon bash

Nicely formatted output of tests (colors):

docker exec -it semicon test -v