Refactor the approach from feature/exact_state so that GeneralState is the top level class, making the API closer to the upcoming version of cuTensorNet and also more similar to the approach in structured_state submodule. In particular:
configure().prepare().compute() from GeneralState are combined into a get_statevector() which will free the scratch space at the end of the computation.
GeneralExpectationValue is also removed and its corresponding chain of methods configure().prepare().compute() is combined into a expectation_value() method within GeneralState.
GeneralOperator is removed since it's too low-level and only relevant for expectation value calculations. The cuTensorNet NetworkOperator only needs to survive during the application of the new method expectation_value().
Refactor the approach from
feature/exact_stateso thatGeneralStateis the top level class, making the API closer to the upcoming version of cuTensorNet and also more similar to the approach instructured_statesubmodule. In particular:configure().prepare().compute()fromGeneralStateare combined into aget_statevector()which will free the scratch space at the end of the computation.GeneralExpectationValueis also removed and its corresponding chain of methodsconfigure().prepare().compute()is combined into aexpectation_value()method withinGeneralState.GeneralOperatoris removed since it's too low-level and only relevant for expectation value calculations. The cuTensorNetNetworkOperatoronly needs to survive during the application of the new methodexpectation_value().