- Fixed bug in :class:`~qsw_mpi.MPI.LQSW.initial_state` whereby argument 'mixed' was not correctly passed to :class:`~qsw_mpi.MPI.walk.initial_state`.
- All :class:`~qsw_mpi.MPI.walk` sub-classes will now initialise \rho(0) as a maximally mixed state on passing 'mixed' to :meth:`~qsw_mpi.MPI.walk.initial_state`.
- Fixed incorrect superoperator generation when using the :class:`~qsw_mpi.MPI.GQSW` class which occurred when a global Lindblad operator had an empty row aligning with the first row of the local partition of the L^TL^* \otimes I_N term in \tilde{\mathcal{L}}.
- Edited the 'Theory' and 'Package Overview' sections of the package documentation to be in line with reviewer comments on the corresponding journal article.
- Changed plot sizes and plot text sizes in response to reviewer feedback.
- Enabled plot legends in desktop benchmarks output.
- Desktop accuracy benchmarks now consider only \mathrm{max}(|\Delta\rho(t)|).
This is a major revision of QSW_MPI. The focus of this release is the expansion of the simulation capabilities of QSW_MPI while focussing the scope of the package through the removal of features which are better supported through pre-existing alternatives (specifically file I/O and visualisation).
- Generalised support for quantum stochastic walks, including the non-moralising quantum stochastic walk through the :class:`~qsw_mpi.MPI.LQSW` and :class:`~qsw_mpi.MPI.GQSW` classes.
- Experimental support for sparse systems following the Gorini–Kossakowski–Sudarshan–Lindblad equation in its diagonalised form through the :class:`~qsw_mpi.MPI.GKSL` class.
- Support for MPI-enabled parallel output to HDF5 using H5Py via the non-user accessible module :mod:`~qsw_mpi.parallel_io`.
- Additional operator types including the canonical Markov chain transition matrix, and those required for the demoralisation correction scheme.
- All simulation types are now subclasses a generalised :class:`~qsw_mpi.MPI.walk` class. This breaks compatibility with version 0.0.1.
- :meth:`~qsw_mpi.MPI.walk.step` and :meth:`~qsw_mpi.MPI.walk.series` have been simplified, gathering of simulation results, or saving of the simulation results is now carried out through the :meth:`~qsw_mpi.MPI.walk.gather_result`, :meth:`~qsw_mpi.MPI.walk.gather_populations`, :meth:`~qsw_mpi.MPI.save_result` or :meth:`~qsw_mpi.MPI.save_populations`.
- Removed visualisation module :mod:`~qsw_mpi.plot`. For basic visualisation, direct use of Matplotlib and Networkx is recommended.
- Removed dedicated I/O module :mod:`~qsw_mpi.io`. For HDF5 file operations, direct use of H5Py is recommended.
Note
This version supports only quantum stochastic walk simulation with locally interacting Lindblad operators.