[Wavepacket-user] Version 0.3.6 released (C++ / Python version)

[Ulf L. <ul...@wp...>]

Hello everyone,

I wanted to announce the next release of the C++ branch of Wavepacket.
This is also the first version that should feature something that the
Matlab version does not offer: Faber polynomials.

Packages can be downloaded under
https://sourceforge.net/projects/cpp.wavepacket.p/files/releases/0.3.6/


The major changes:

* A Windows binary package of the Python interface.

  After a lot of wrangling, I finally managed to put together something
  that you hopefully can just import in Python under Windows. The
  package is still crude, and the documentation mainly explains the C++
  code (which mostly maps rather easily to Python), but this should
  allow Windows users to use the library as well.

  Along the way, various other changes have been implemented. I updated
  the tensor dependency, and there is now something closer to a
  compilation Howto in the form of a Vagrantfile that tries to build
  all dependencies from scratch.


* 2D contour plots for the Python interface.

  They are still inferior to the Matlab plots, but at least you can
  principally visualize 2D systems now. See
  https://sourceforge.net/p/wavepacket/wiki/Demos.HarmOscillator.Gaussian2D/


* A solver based on Faber polynomials.

  Faber polynomials are a generalization of Chebychev polynomials that
  work for open quantum systems / complex Hamiltonians. They retain
  many advantages of the Chebychev solver, in particular efficiency,
  but are more difficult to use. For more details, see the highlighted
  demo:
  
  https://wavepacket.sourceforge.io/cpp-doc/current/demo_faber.html

  This feature also has some spin-offs. Because the solver is not
  useful without good spectrum estimation and truncation, I added some
  functionality to get a minimum/maximum potential, estimate an
  eigenvalue with the power method, and for truncating certain simple
  operators (diagonal in DVR or FBR).


As usual, there are also a bunch of smaller changes, which you can read
in the full changelog:
https://sourceforge.net/p/wavepacket/cpp/git/ci/master/tree/NEWS.


For the next releases, I would focus on implementing missing
functionality to close up to the Matlab version, and on getting the API
into the final shape. This will unfortunately require some breaking
changes here and there.


Regards,
Ulf