Quobly forge emulator

The Quobly-alloy SDK adresses Quobly QPUs such as Alloy Pioneer and their emulators using the forge module.

Installation

You can install Quobly Alloy with

pip install quobly-alloy

Use

The minimal program to call an emulator such as the Pioneer emulator is

from qiskit import QuantumCircuit
from quobly_alloy.forge import PioneerEmulator
from quobly_alloy import QPU

circuit = QuantumCircuit(2)
circuit.rx(1, 1)
circuit.rz(1, 0)
circuit.rx(1, 0)
circuit.rzz(1, 0, 1)
circuit.measure_all()

emulator = PioneerEmulator(QPU.PIONEER_P10)
result = emulator.run_simulation(circuit,shots=1000)
print(result)

This code first create a circuit of 2 qubits, then simulate it on the PIONEER_P10 machine using run simulation. The methods PioneerEmualtor.run_simulation simulate a circuit for one ten shots. This return a dictionary[str,int] composed of key being the bitstring of the machine and values being the number of time the bitstring appears.

One can also use the function run, that return a QuoblyJob object (inheriting from Qiskit.Job) with the methods QuoblyJob.result that return the same result as run_simulation. This methods exist for adherence to qiskit framework.

Furthermore, one can fix a seed using

emulator = PioneerEmulator(QPU.PIONEER_P10,seed = 100)
result = emulator.run_simulation(circuit=circuit,shots=1000)

In addition one can select the number of core used for the emulation with (note that max_used_core is capped by the number of core of your hardware):

emulator = PioneerEmulator(QPU.PIONEER_P10,max_used_core=6)
result = emulator.run_simulation(circuit=circuit,shots=1000,noise=False)

You can also remove the injected noise using:

emulator = PioneerEmulator(QPU.PIONEER_P10)
result = emulator.run_simulation(circuit=circuit,shots=1000,noise=False)

Finally you can change the number of qubits using:

emulator = PioneerEmulator(QPU.PIONEER_P10,qubits = 5)
result = emulator.run_simulation(circuit=circuit)

[!CAUTION] The number of possible qubits is dependant of the computer memory size.

Transpilation

The transpilation is done internally by the backend.

Using qiskit-aer-gpu for cuda 12

As of no, qiskit 2.x does not support cuda 12, but conda does.

On your conda environment you can install the correct package using

conda install -c conda-forge "qiskit-aer=0.17.2=*cuda*" cuda-version=12

You can then check if the gpu is correctly found using:

python -c "from qiskit_aer import AerSimulator; print('GPU' in AerSimulator().available_devices())"

You can then use the backend as normal, the backend prioritize the GPU device if found. You can force the use of the CPU using:

emulator = PioneerEmulator(QPU.PIONEER_P10,always_use_cpu = True)