A library to use Qiskit Transpiler service and the AI transpiler passes.
Note The Qiskit transpiler service and the AI transpiler passes use different experimental services that are only available for IBM Quantum Premium Plan users. This library and the releated services are an alpha release, subject to change.
To use the Qiskit transpiler service, install the qiskit-transpiler-service
package:
pip install qiskit-transpiler-service
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 library requires Qiskit 1.0 by default.
The following examples demonstrate how to transpile circuits using the Qiskit transpiler service with different parameters.
-
Create a circuit and call the Qiskit transpiler service to transpile the circuit with
ibm_sherbrooke
as thebackend_name
, 3 as theoptimization_level
, and not using AI during the transpilation.from qiskit.circuit.library import EfficientSU2 from qiskit_transpiler_service.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 are allowed to with your IBM Quantum Account. Apart from the backend_name
, the TranspilerService
also allows coupling_map
as parameter.
-
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_transpiler_service.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)
The AIRouting
pass acts both as a layout stage and a routing stage. It can be used within a PassManager
as follows:
from qiskit.transpiler import PassManager
from qiskit_transpiler_service.ai.routing import AIRouting
from qiskit.circuit.library import EfficientSU2
ai_passmanager = PassManager([
AIRouting(backend_name="ibm_sherbrooke", optimization_level=2, layout_mode="optimize")
])
circuit = EfficientSU2(101, entanglement="circular", reps=1).decompose()
transpiled_circuit = ai_passmanager.run(circuit)
Here, the backend_name
determines which backend to route for, the optimization_level
(1, 2, or 3) determines the computational effort to spend in the process (higher usually gives better results but takes longer), and the layout_mode
specifies how to handle the layout selection.
The layout_mode
includes the following options:
keep
: This respects the layout set by the previous transpiler passes (or uses the trivial layout if not set). It is typically only used when the circuit must be run on specific qubits of the device. It often produces worse results because it has less room for optimization.improve
: This uses the layout set by the previous transpiler passes as a starting point. It is useful when you have a good initial guess for the layout; for example, for circuits that are built in a way that approximately follows the device's coupling map. It is also useful if you want to try other specific layout passes combined with theAIRouting
pass.optimize
: This is the default mode. It works best for general circuits where you might not have good layout guesses. This mode ignores previous layout selections.
The AI circuit synthesis passes allow you to optimize pieces of different circuit types (Clifford, Linear Function, Permutation) by re-synthesizing them. The typical way one would use the synthesis pass is the following:
from qiskit.transpiler import PassManager
from qiskit_transpiler_service.ai.routing import AIRouting
from qiskit_transpiler_service.ai.synthesis import AILinearFunctionSynthesis
from qiskit_transpiler_service.ai.collection import CollectLinearFunctions
from qiskit.circuit.library import EfficientSU2
ai_passmanager = PassManager([
AIRouting(backend_name="ibm_cairo", optimization_level=3, layout_mode="optimize"), # Route circuit
CollectLinearFunctions(), # Collect Linear Function blocks
AILinearFunctionSynthesis(backend_name="ibm_cairo") # Re-synthesize Linear Function blocks
])
circuit = EfficientSU2(10, entanglement="full", reps=1).decompose()
transpiled_circuit = ai_passmanager.run(circuit)
The synthesis respects the coupling map of the device: it can be run safely after other routing passes without "messing up" the circuit, so the overall circuit will still follow the device restrictions. By default, the synthesis will replace the original sub-circuit only if the synthesized sub-circuit improves the original (currently only checking CNOT
count), but this can be forced to always replace the circuit by setting replace_only_if_better=False
.
The following synthesis passes are available from qiskit_transpiler_service.ai.synthesis
:
- AICliffordSynthesis: Synthesis for Clifford circuits (blocks of
H
,S
andCX
gates). Currently up to 9 qubit blocks. - AILinearFunctionSynthesis: Synthesis for Linear Function circuits (blocks of
CX
andSWAP
gates). Currently up to 9 qubit blocks. - AIPermutationSynthesis: Synthesis for Permutation circuits (blocks of
SWAP
gates). Currently available for 65, 33, and 27 qubit blocks.
We expect to gradually increase the size of the supported blocks.
All passes use a thread pool to send several requests in parallel. By default it will use as max threads as number of cores plus four (default values for ThreadPoolExecutor
python object). However, you can set your own value with the max_threads
argument at pass instantation. For example, the following line will instantiate the AILinearFunctionSynthesis
pass allowing it to use a maximum of 20 threads.
AILinearFunctionSynthesis(backend_name="ibm_cairo", max_threads=20) # Re-synthesize Linear Function blocks using 20 threads max
You can also set the environment variable AI_TRANSPILER_MAX_THREADS
to the desired number of maximum threads, and all synthesis passes instantiated after that will use that value.
For sub-circuit to be synthesized by the AI synthesis passes, it must lay on a connected subgraph of the coupling map (this can be ensured by just doing a routing pass previous to collecting the blocks, but this is not the only way to do it). The synthesis passes will automatically check if a the specific subgraph where the sub-circuit lays is supported, and if it is not supported it will raise a warning and just leave the original sub-circuit as it is.
To complement the synthesis passes we also provide custom collection passes for Cliffords, Linear Functions and Permutations that can be imported from qiskit_transpiler_service.ai.collection
:
- CollectCliffords: Collects
Clifford
blocks asInstruction
objects and stores the original sub-circuit to compare against it after synthesis. - CollectLinearFunctions: Collects blocks of
SWAP
andCX
asLinearFunction
objects and stores the original sub-circuit to compare against it after synthesis. - CollectPermutations: Collects blocks of
SWAP
circuits asPermutations
.
These custom collection passes limit the sizes of the collected sub-circuits so that they are supported by the AI synthesis passes, so it is recommended to use them after the routing passes and before the synthesis passes to get a better optimization overall.
The library is prepared to let the user log the messages they want. For that, users only have to add the following code to their code:
import logging
logging.getLogger("qiskit_transpiler_service").setLevel(logging.X)
where X can be: NOTSET
, DEBUG
, INFO
, WARNING
, ERROR
or CRITICAL
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},
}