The (Qiskit)-token should not be stored in the DB.
Currently, it is required for the compilation (send via endpoints: selection/compiler-selection/qpu-selection ...) and at the same time stored in the DB for the later execution of analysis/compilation/qpu-selection results. Both via Qiskit-/Pytket-/Forest-Service (see Connectors in NISQ Analyzer).
It is actually (at least) stored in these tables:

• analysis_job_input_parameters
• analysis_result_input_parameters
• compilation_result
• qpu_selection_result

Instead, the token should not only be sent over the root endpoints (selection/compiler-selection/qpu-selection), but should also be sent over the corresponding execution-endpoints that also require the token to remove the DB dependency.
Therefore, first check again in which cases the token is currently taken out of the DB for processing.
Then, remove the token out of the respective tables and add it to the respective endpoints.

Note: Thereby, refreshToken or bearerToken is NOT meant as it is not stored in the DB - instead, the IBMQ token is meant, required to start an analysis, see, e.g., screenshot.

Note: Also the UI has to be updated, such that the token is added to the execution requests.

after /algorithm/id/nisq-analyzer/selection show all suitable implementations not only one

For integrating the implementations of the PlanQK Platform for analysis, the NISQ Analyzer

• provide API for QPU selection for a specific implementation (with input parameters)
• transfer bearer token of Platform user to Com&Exe Services (qiskit-service, pytket-service, forest-service)

Implemented functionalities have to be removed:

• support estimatedDepth and estimatedWidth rules

Acceptance Criteria:

1. remove expectedDepth, expectedWidth rules
2. remove pre-filtering with expectedDepth, expectedWidth values
3. use "Constants.QPU_RULE_CONTENT" for QPU selection for transpiled values
4. remove estimated flag

Create a docker-compose that contains nisq-analyzer and qiskit-service.

Acceptance Criteria:

1. docker-compose containing nisq-analyzer, qiskit-service, qc-atlas-ui, qc-atlas

The credentials for the provider should not be saved in the provider class. They should be passed through to the Qiskit Service.

Acceptance criteria:

1. Not save the credentials in the provider class
2. credentials should be handed over to the Qiskit Service
3. Single token or username and password should be possible

In context of Algorithm > Implementation > NISQ Analyzer:
When starting a new analysis, the user should be able to select what SDKs/compilers should be considered for compilation and analysis in the pop-up.
In case the cloud provider is IBMQ, pytket and Qiskit should be selectable (and shown).
In case the cloud provider is Rigetti, pytket and Forest should be selectable (and shown).

Note: At least one of the presented SDKs has to be selected, otherwise the analysis can not start.

Based on the selection, the compilersToUse list should be set.

Implement service to use t|ket> compiler aka pytket (such as qiskit-service) and implement connector for NISQ Analyzer

Acceptance Criteria

1. implement pytket service
2. implement connector in nisq-analyzer

• Update old tests
• add new ones
• define GitHub Actions Workflow for automated execution of tests
• for integration tests start docker containers of SDK Service, QuantumTranspiler, QProv, etc. with latests images in the workflow

NISQ Analyzer has to support a regex library for the implementation selection rules.

Acceptance Criteria:

1. Add "pack_install(regex)" to Dockerfile
2. Add ":- use_module(library(regex))." to each generated .pl file

See: https://www.swi-prolog.org/pack/list?p=regex

Beneath execution results of transpilation services also save date and time.

Remove nisq-analyzer from the path, find better path naming.

Acceptance Criteria:

1. nisq-analyzer/algorithms/id/nisq-analyzer/selection --> nisq-analyzer/algorithms/id/selection

If prolog rules estimate, that an implementation is executable but the transpiler (qiskit-service) cannot transpile it because too many qubits are required, the "estimate" flag is false but analysedDepth and analysedWidth is 0.

Therefore, this case should not be shown in response.

After restart prolog files of implementations and qpus that are already saved in the DB should be generated again.

Acceptance Criteria

1. If implementations, qpus are already saved in db: after restart selection and execution should be possible without having to create them again.

Example:
Request-Body with String-parameter in nisq-analyzer/selection is "Oracle": "'(A | B) & (A | ~B) & (~A | B)'"
' ' required for prolog rules.
Qiskit-Service cannot parse "'(A | B) & (A | ~B) & (~A | B)'" with ' '.
Therefore, for String-parameters the ' ' must be deleted before sending them to the service.

Currently JPL version is too old for Prolog versions => 8.0.
But, currently, maven seems to not support the latest JPL version!
Please find out if there is, currently, really no way.