Description
Environment
- Qiskit Machine Learning version: 0.8.2
- Python version: 3.10
- Operating system: Ubuntu 24.04.2 LTS
- qiskit:1.4.2
- qiskit-ibm-runtime:0.37.0
What is happening?
When I try to run a SamplerQNN on IBM Quantum hardware (ibm_kyiv in my case), I get the following error:
QiskitMachineLearningError: "Sampler job failed: 'Unable to retrieve job result. Error code 3211; Job not valid. list index out of range'"
The error also shows up on the IBM Quantum platform job page.
Interestingly, the code works without issues when I use a local or GenericBackendV2 simulator. So the problem seems specific to the IBM hardware or its interaction with SamplerV2 and SamplerQNN.
How can we reproduce the issue?
To evaluate the error, run the following minimal working example using an IBM Quantum backend such as ibm_kyiv. Make sure to replace "MY_TOKEN" with an IBM Quantum token.
import numpy as np
from qiskit.circuit.library import RealAmplitudes, ZZFeatureMap
from qiskit.circuit import QuantumCircuit
from qiskit_machine_learning.neural_networks import SamplerQNN
from qiskit_ibm_runtime import QiskitRuntimeService
from qiskit_ibm_runtime import SamplerV2 as Sampler
from qiskit.transpiler.preset_passmanagers import generate_preset_pass_manager
from qiskit_machine_learning.gradients import ParamShiftSamplerGradient
service1 = QiskitRuntimeService(channel='ibm_quantum', token='MY_TOKEN')
backend = service1.backend('ibm_kyiv')
pass_manager = generate_preset_pass_manager(optimization_level=0, backend=backend)
sampler = Sampler(mode=backend)
sampler_gradient = ParamShiftSamplerGradient(sampler=sampler, pass_manager=pass_manager)
num_qubits = 2
def parity(x):
return f"{bin(x)}".count("1") % 2
feature_map = ZZFeatureMap(feature_dimension=num_qubits)
ansatz = RealAmplitudes(num_qubits=num_qubits)
qc = QuantumCircuit(num_qubits)
qc.compose(feature_map, inplace=True)
qc.compose(ansatz, inplace=True)
qnn = SamplerQNN(
circuit=qc,
input_params=feature_map.parameters,
weight_params=ansatz.parameters,
interpret=parity,
output_shape=2,
sampler=sampler,
pass_manager=pass_manager,
)
result = qnn.forward(input_data=[1, 2], weights=[1, 2, 3, 4, 5, 6, 7, 8])
print(result)
This throws the following traceback:
---------------------------------------------------------------------------
RuntimeJobFailureError Traceback (most recent call last)
File ~/anaconda3/envs/fraconda/lib/python3.10/site-packages/qiskit_machine_learning/neural_networks/sampler_qnn.py:497, in SamplerQNN._forward(self, input_data, weights)
496 try:
--> 497 results = job.result()
498 except Exception as exc:
File ~/anaconda3/envs/fraconda/lib/python3.10/site-packages/qiskit_ibm_runtime/runtime_job_v2.py:137, in RuntimeJobV2.result(self, timeout, decoder)
136 raise RuntimeJobMaxTimeoutError(error_message)
--> 137 raise RuntimeJobFailureError(f"Unable to retrieve job result. {error_message}")
138 if self._status == "CANCELLED":
RuntimeJobFailureError: 'Unable to retrieve job result. Error code 3211; Job not valid. list index out of range'
The above exception was the direct cause of the following exception:
QiskitMachineLearningError Traceback (most recent call last)
Cell In[10], line 29
17 qnn = SamplerQNN(
18 circuit=qc, # Note that this is a QuantumCircuit instance
19 input_params=feature_map.parameters,
(...)
24 pass_manager=pass_manager,
25 )
28 # Do a forward pass with input data and custom weights
---> 29 result=qnn.forward(input_data=[1, 2], weights=[1, 2, 3, 4, 5, 6, 7, 8])
30 print(result)
File ~/anaconda3/envs/fraconda/lib/python3.10/site-packages/qiskit_machine_learning/neural_networks/neural_network.py:229, in NeuralNetwork.forward(self, input_data, weights)
227 input_, shape = self._validate_input(input_data)
228 weights_ = self._validate_weights(weights)
--> 229 output_data = self._forward(input_, weights_)
230 return self._validate_forward_output(output_data, shape)
File ~/anaconda3/envs/fraconda/lib/python3.10/site-packages/qiskit_machine_learning/neural_networks/sampler_qnn.py:499, in SamplerQNN._forward(self, input_data, weights)
497 results = job.result()
498 except Exception as exc:
--> 499 raise QiskitMachineLearningError(f"Sampler job failed: {exc}") from exc
500 result = self._postprocess(num_samples, results)
501 return result
QiskitMachineLearningError: "Sampler job failed: 'Unable to retrieve job result. Error code 3211; Job not valid. list index out of range'"
What should happen?
The SamplerQNN should successfully run the forward pass and return output probabilities (or other configured values) instead of throwing an exception. Since the same code works with a generic backend, this seems to be an integration issue when running on real IBM hardware.
Any suggestions?
No response