Environment
- povm-toolbox version: 0.2.0
- Python version: 3.14.4
- Operating system: Mac OS
What is happening and why is it wrong?
When I submit a batch job of more than 81 PUBs, the job fails with the error (taken from the IBM Quantum Platform website):
2026/04/23 02:21:52.625 - user_logger - ERROR Error executing program for job_id d7ko4b8kj84c73ce4ug0: Internal Compiler Error -- https://ibm.biz/error_codes#6053:
2026/04/23 02:21:52.684 - user_logger - ERROR Failed to execute program: Internal Compiler Error -- https://ibm.biz/error_codes#6053
How can we reproduce the issue?
A minimum breaking example seems to be:
from qiskit import QuantumCircuit
from qiskit_ibm_runtime import QiskitRuntimeService
from qiskit_ibm_runtime import SamplerV2 as Sampler
from qiskit.transpiler import generate_preset_pass_manager
from povm_toolbox.sampler import POVMSampler
from povm_toolbox.library import ClassicalShadows
from povm_toolbox.post_processor import POVMPostProcessor
backend = service.least_busy(operational=True, simulator=False)
qc = QuantumCircuit(2)
qc.h(0)
qc.cx(0,1)
qcs = [qc]*82
pm = generate_preset_pass_manager(backend=backend, optimization_level=3)
isa_qcs = pm.run(qcs)
sampler = Sampler(mode=backend)
povm_sampler = POVMSampler(sampler=sampler)
measurement = ClassicalShadows(num_qubits=n, seed=42)
job = povm_sampler.run(isa_qcs, shots=1000, povm=measurement)
print(f">>> Job ID: {job.job_id()}")
This just sends 82 Bell state prep circuits to the device (Fez in my case, but the error occurs on other devices too). If the length of qcs is 81, the error disappears. The number of shots does not seem to have any effect one way or another. Using an Estimator to measure some 2-qubit observable instead of the shadows routine also resolves the error. So there appears to be some weird limit only when using this toolbox?
Traceback
No response
Any suggestions?
No response
Environment
What is happening and why is it wrong?
When I submit a batch job of more than 81 PUBs, the job fails with the error (taken from the IBM Quantum Platform website):
How can we reproduce the issue?
A minimum breaking example seems to be:
This just sends 82 Bell state prep circuits to the device (
Fezin my case, but the error occurs on other devices too). If the length ofqcsis 81, the error disappears. The number of shots does not seem to have any effect one way or another. Using anEstimatorto measure some 2-qubit observable instead of the shadows routine also resolves the error. So there appears to be some weird limit only when using this toolbox?Traceback
No response
Any suggestions?
No response