Pygsti

Added

- adds a new API which centers around Protocol objects, and makes it easier to run one or multiple protocols on a many-qubit quantum processor. See the new 00-Protocols.ipynb tutorial for more details.
- new benchmarking protocols, including volumetric benchmarks and mirror randomized-benchmarking.
- more gauge-invariant metrics are available within reports.

Changed

- improvements to multi-qubit GST capabilities.
- performance improvements, particularly in the area of working with large data-sets.
- HTML reports have been updated to by default use a new AJAX-free framework that eliminates many of the browser compatibility problems that users experienced in the past.
- Python version 2 support has been dropped. Python 3.5 and higher are now the only python versions pyGSTi officially supports.\
- the root package directory has been updated for greater simplicity and standardization with other packages. If you "install" new pyGSTi versions by running "git pull", you'll also need to run pip -e install . again or you'll get a warning message the pyGSTi cannot be found.

Fixed

- Fixed the issue of weird ('\*',) line labels in circuit construction routines when used with multi-qubit notation circuits, e.g. (0,)-labeled circuits getting erroneous ('\*',0)-labeled circuits created.
- Fixed bug in optools.fidelity for special case of trace-reducing rank-1 ops.
- Now handling sqrtm failure in optools._hack_sqrt as a warning instead of assertion error.

Fixed

- adds compensation for a scaling factor in matrixtools.unitary_superoperator_matrix_log which would cause an assertion to fail in some circumstances.

Changed

- pyGSTi is distributed with the Apache License version 2.0, included in LICENSE.

Added

- framework for time-dependent models and continuously-parameterized gates.
- added support for simple Pauli-stochastic noise operators.
- new tutorials for: idle tomgraphy, drift characterization, time-dependent GST, simulating RB data, continuously-parameterized gates, building implicit models (updated with crosstalk-free and cloud-crosstalk constructions)

Changed

- revamped drift detection code
- expanded Circuit objects can include arguments and time durations.
- expanded ability to create implicit models (local- and cloud-noise models) makes this process easier and more intuitive. Related to this, the error generators of LindbladOp objects can more easily be constructed and modified.
- more robust support for different bases (adds a new hierarchy of basis objects).
- updated documentation and bug fixes

Fixed

- adds the package pygsti.extras.idletomography which was missing from setup.py.