wavepacket 0.2.0

A package for the propagation of quantum-mechanical wave functions.

Description

Wavepacket is a Python package to define and simulate small quantum systems. Or, more technically, it allows you to numerically solve Schrödinger and Liouville-von-Neumann equations for distinguishable particles.

The full documentation can be found under https://wavepacket.readthedocs.io.

There are many different quantum systems and consequently approaches to solve them. Here we focus on a particular niche:

• Wavepacket solves the differential equations directly. This simplifies the maths, but limits the system size to few degrees of freedom. If you want to deal with larger systems, look out for MCTDH.

• Wavepacket uses the DVR approximation heavily. This allows you to directly define your potentials as functions of real-space coordinates instead of setting up opaque operator matrices. The latter approach is simpler an more concise if you are only interested in harmonic oscillators or qubits, though.

• Wavepacket is a Python-only package relying chiefly on numpy. This is slower than natively implemented code, but you gain great tooling support, for example matplotlib, Jupyter notebooks or integrated documentation.

• Most of the code can handle both wave functions and density operators. This allows you to convert a closed system into an open system with minimal fuss.

For example use cases, we have been using various precursors of this package for simulating small molecular systems and for teaching. Besides examples shipped with this package, see https://sourceforge.net/p/wavepacket/wiki/Demos.Main for more applications.

The project is currently a first iteration to flesh out everything. Once 0.1 is released, I plan to quickly translate the existing C++ code from a precursor project and reach a stable state. More can be found on the project homepage https://github.com/ulflor/wavepacket

Support

If you lack a feature that you would like to have, open an issue at our issue tracker. Depending on the complexity of the feature, this will lead to an immediate, rapid, or prioritized implementation.

Contribution

I currently lack a formal procedure for new contributors, but you are very welcome to contribute to the project. If you do not know what to do, feel free to contact one of the developers; there is enough work for multiple developers, it is just not documented yet.

History

The original version of Wavepacket was written in Matlab and is still maintained under https://sourceforge.net/p/wavepacket/matlab. It is stable, battle-tested and works.

However, Matlab is pretty expensive, so not all interested users had access to it. Also, the project’s architecture did not support some advanced use cases without digging deep into the code. Finally, C++11 had just come out and looked cool, so I started a reimplementation in C++ around 2013, adding Python bindings as an afterthought. The C++ project will be superseded by this Python package, but can be found under https://sourceforge.net/p/wavepacket/cpp.

This worked really well. However, deploying C++ code is difficult. In particular, there was no cheap route towards building a “good” Python package, or towards easily building a Windows version. Also, the underlying tensor library was slowly getting fewer and fewer commits over the years, so I am currently moving to a Python-only package. The Python version is slower by a factor of two to three compared to C++-backed code. This is, however, often cancelled by a parallelization of the tensor operations, and the tooling is better by orders of magnitude.