Additions & Improvements
---------------------------------
- Added functions `real` and `imag` to `cplx` sub-module
- Negative Log Likelihood has been added to the `training_statistics` submodule.
- Pauli observables can now optionally compute the absolute magnetization of a chain
- Can now get data from `MetricEvaluator` and `ObservableEvaluator` using list/dictionary subscripting syntax. See the "Training" section of Tutorial 1 for more details.
Fixes
-------
- `RenyiEntropy` observable has been replaced by `SWAP` observable, as the old implementation produced biased samples. See the corresponding section of Tutorial 3 for more details on how to use the `SWAP` observable to compute the 2nd Renyi Entropy.
Misc
------
- Refined dependency requirements:
- Maximum supported PyTorch version set to 1.2 (doesn't exist yet, restriction exists just for safety). Minimum version is still 1.0
- Latest version of matplotlib no longer supports Python 3.5, thus, the maximum supported matplotlib version for systems running Python 3.5 is version 3.1. Minimum version for all supported Python versions is still 2.2