Clean-up

[eggerdj]

What should we add?

We need to clean-up the transpilation part of the best-practices. This involves making the annotated transpilation the default. Ideally, from the user's perspective the workflow should look like the following (in pseudo code):

cost_op = ...  # cost-operator obtained from an optimization modelling framework

# SAT mapping to pre-process the cost operator
sat_mapper = SATMapper()
mapped_op = sat_mapper.remap_with_sat(cost_op, swap_strategy)

# Qubit selection based on a subclass of BackendEvaluator
qubit_evaluator = ChildOfBackendEvaluator(backend)
qubit_layout = qubit_evaluator.find(mapped_op)

# Prepare the transpilation pass
config = {
    "initial_layout": qubit_layout,
    "swap_strategy": swap_strategy,
}
pm = preset_qaoa_passmanager(backend, **config)

ansatz = annotated_qaoa_ansatz(cost_op, reps=2)

isa_ansatz = pm.run(ansatz)  # A hardware ready circuit

This objective defines the following set of TODOs:

• Allow the SAT mapper to accept cost operators as input. (See PR Refactor the qaoa preset pass manager #48 )
• Create the function preset_qaoa_passmanager by refactoring qaoa_swap_strategy_pm in preset_qaoa_passmanager.py (See PR Refactor the qaoa preset pass manager #48 )
• Allow more flexibility in the qubit evaluation and support more options (e.g. HW native)
• Make the annotated functionality centre stage by giving the different classes their own files.
• Make it clear the the pauli-evolution is for older Qiskit version.