Transpile circuits remotely with the Qiskit Transpiler Service

The Qiskit Transpiler Service provides transpilation capabilities on the cloud. In addition to the local Qiskit transpiler capabilities, your transpilation tasks can benefit from both IBM Quantum™ cloud resources and AI-powered transpiler passes.

The Qiskit Transpiler Service offers a Python library to seamlessly integrate this service and its capabilities into your current Qiskit patterns and workflows.

Install the qiskit-ibm-transpiler package

To use the Qiskit Transpiler Service, install the qiskit-ibm-transpiler package:

pip install qiskit-ibm-transpiler

By default, the package tries to authenticate to IBM Quantum services with the defined Qiskit API token, and uses your token from the QISKIT_IBM_TOKEN environment variable or from the file ~/.qiskit/qiskit-ibm.json (under the section default-ibm-quantum).

Note: This package requires Qiskit SDK v1.X.

qiskit-ibm-transpiler transpile options

• backend_name (optional, str) - A backend name as it would be expected by QiskitRuntimeService (for example, ibm_sherbrooke). If this is set, the transpile method uses the layout from the specified backend for the transpilation operation. If any other option is set that impacts these settings, such as coupling_map, the backend_name settings are overridden.
• coupling_map (optional, List[List[int]]) - A valid coupling map list (for example, [[0,1],[1,2]]). If this is set, the transpile method uses this coupling map for the transpilation operation. If defined, it overrides any value specified for target.
• optimization_level (int) - The potential optimization level to apply during the transpilation process. Valid values are [1,2,3], where 1 is the least optimization (and fastest), and 3 the most optimization (and most time-intensive).
• ai ("true", "false", "auto") - Whether to use AI-powered capabilities during transpilation. The AI-powered capabilities available can be for AIRouting transpiling passes or other AI-powered synthesis methods. If this value is "true", the service applies different AI-powered transpiling passes depending on the optimization_level requested. If "false", it uses the latest Qiskit transpiling features without AI. Finally, if "auto", the service decides whether to apply the standard Qiskit heuristic passes or the AI-powered passes based on your circuit.
• qiskit_transpile_options (dict) - A Python dictionary object that can include any other option that is valid in the Qiskit transpile() method. If the qiskit_transpile_options input includes optimization_level, it is discarded in favor of the optimization_level specified as parameter input. If the qiskit_transpile_options includes any option not recognized by the Qiskit transpile() method, the library raises an error.

For more information about the available qiskit-ibm-transpiler methods, see the qiskit-ibm-transpiler API reference. To learn more about the service API, see the Qiskit Transpiler Service REST API documentation.

Examples

The following examples demonstrate how to transpile circuits using the Qiskit Transpiler Service with different parameters.

1. Create a circuit and call the Qiskit Transpiler Service to transpile the circuit with ibm_sherbrooke as the backend_name, 3 as the optimization_level, and without using AI during the transpilation.

   from qiskit.circuit.library import EfficientSU2
   from qiskit_ibm_transpiler.transpiler_service import TranspilerService
    
   circuit = EfficientSU2(101, entanglement="circular", reps=1).decompose()
    
   cloud_transpiler_service = TranspilerService(
       backend_name="ibm_sherbrooke",
       ai="false",
       optimization_level=3,
   )
   transpiled_circuit = cloud_transpiler_service.run(circuit)

Note: you only can use backend_name devices you have access to with your IBM Quantum account. Apart from the backend_name, the TranspilerService also allows coupling_map as parameter.

2. Produce a similar circuit and transpile it, requesting AI transpiling capabilities by setting the flag ai to True:

   from qiskit.circuit.library import EfficientSU2
   from qiskit_ibm_transpiler.transpiler_service import TranspilerService
    
   circuit = EfficientSU2(101, entanglement="circular", reps=1).decompose()
    
   cloud_transpiler_service = TranspilerService(
       backend_name="ibm_sherbrooke",
       ai="true",
       optimization_level=1,
   )
   transpiled_circuit = cloud_transpiler_service.run(circuit)

3. Produce a similar circuit and transpile it while letting the service to decide whether to use the AI-powered transpiling passes.

   from qiskit.circuit.library import EfficientSU2
   from qiskit_ibm_transpiler.transpiler_service import TranspilerService
    
   circuit = EfficientSU2(101, entanglement="circular", reps=1).decompose()
    
   cloud_transpiler_service = TranspilerService(
       backend_name="ibm_sherbrooke",
       ai="auto",
       optimization_level=1,
   )
   transpiled_circuit = cloud_transpiler_service.run(circuit)

Limits of the Qiskit Transpiler Service

Following are the most relevant limitations of the service:

• The maximum number of two-qubit gates per circuit in a transpilation job in any ai mode is 1 million.
• The maximum time allowed to run a transpilation process is 30 minutes per job.
• You must retrieve the transpilation result from the service within 20 minutes after the transpilation process ends. After 20 minutes, the job result is discarded.
• The maximum time a set of circuits can live in the internal queue while waiting to be transpiled is 120 minutes. After that time, if the job has not been transpiled, it is discarded.
• The maximum number of qubits has not been determined. The service has been tested on 900+ qubits.

Citation

If you use any AI-powered feature from the Qiskit Transpiler Service in your research, use the following recommended citation:

@misc{2405.13196,
Author = {David Kremer and Victor Villar and Hanhee Paik and Ivan Duran and Ismael Faro and Juan Cruz-Benito},
Title = {Practical and efficient quantum circuit synthesis and transpiling with Reinforcement Learning},
Year = {2024},
Eprint = {arXiv:2405.13196},
}

Next steps