Error when drawing circuits generated by StandardRB

[eendebakpt]

Informations

• Qiskit Experiments version: 0.4.0
• Python version: 3.10
• Operating system: Windows

What is the current behavior?

Running this script:

import qiskit
from qiskit_experiments.library import StandardRB
import qiskit_experiments
import numpy as np
print(qiskit.version.QiskitVersion())
print(qiskit_experiments.version.get_version_info())


lengths = np.arange(1, 600, 200)
exp1 = StandardRB([0], lengths, num_samples=4, seed=1010)
circuits=exp1.circuits()
circuits[0].draw()

results in an exception

{'qiskit-terra': '0.22.2', 'qiskit-aer': '0.11.1', 'qiskit-ignis': None, 'qiskit-ibmq-provider': '0.19.2', 'qiskit': None, 'qiskit-nature': None, 'qiskit-finance': None, 'qiskit-optimization': None, 'qiskit-machine-learning': None}
0.4.0

  ...

  File "C:\develop\env310\lib\site-packages\qiskit\visualization\circuit\_utils.py", line 556, in slide_from_left
    if node.op.condition:

AttributeError: 'Clifford' object has no attribute 'condition'

Steps to reproduce the problem

See the minimal example above

What is the expected behavior?

The circuit should be plotted

Suggested solutions

[conradhaupt]

I have confirmed that this bug occurs with Qiskit Experiments 0.4.0 and Qiskit Terra 0.22.0 0.22.2 on my Mac. Using the current main of both Qiskit Experiments and Qiskit Terra, I get the output below. Is this the output you would have expected? If yes, then I would say this bug is fixed for the next release.

{'qiskit-terra': '0.23.0', 'qiskit-aer': '0.11.0', 'qiskit-ignis': None, 'qiskit-ibmq-provider': '0.19.2', 'qiskit': '0.38.0', 'qiskit-nature': None, 'qiskit-finance': None, 'qiskit-optimization': None, 'qiskit-machine-learning': None}
0.5.0.dev0+d28a4ab
       ░ ┌─────────────────┐ ░ »
   q: ─░─┤ Clifford-1Q(16) ├─░─»
       ░ └─────────────────┘ ░ »
meas: ═════════════════════════»
                               »
«      ┌──────────────────────────────────────────────────────┐ ░ ┌─┐
«   q: ┤ Clifford: Stabilizer = ['-Z'], Destabilizer = ['+Y'] ├─░─┤M├
«      └──────────────────────────────────────────────────────┘ ░ └╥┘
«meas: ════════════════════════════════════════════════════════════╩═
«                                                                    

Edit 2022.11.16 15:42 UTC:

I can confirm that the main version of Qiskit Experiments and Qiskit Terra 0.22.2 works.

{'qiskit-terra': '0.22.2', 'qiskit-aer': None, 'qiskit-ignis': None, 'qiskit-ibmq-provider': '0.19.2', 'qiskit': None, 'qiskit-nature': None, 'qiskit-finance': None, 'qiskit-optimization': None, 'qiskit-machine-learning': None}
0.5.0.dev0+338a46d
       ░ ┌─────────────────┐ ░ »
   q: ─░─┤ Clifford-1Q(16) ├─░─»
       ░ └─────────────────┘ ░ »
meas: ═════════════════════════»
                               »
«      ┌──────────────────────────────────────────────────────┐ ░ ┌─┐
«   q: ┤ Clifford: Stabilizer = ['-Z'], Destabilizer = ['+Y'] ├─░─┤M├
«      └──────────────────────────────────────────────────────┘ ░ └╥┘
«meas: ════════════════════════════════════════════════════════════╩═
«                                                                    

[conradhaupt]

Doing a git bisect from main to 0.4.0 showed that the bug was “fixed” in qiskit-community/qiskit-experiments#898.

[itoko]

This is a bug in Terra. Let's export it as a Terra issue. The following is a minimal code to fail.

from qiskit.circuit import QuantumCircuit
from qiskit.quantum_info.random import random_clifford

circ = QuantumCircuit(1)
circ.append(random_clifford(1), [0])
circ.append(random_clifford(1), [0]) # at least two Clifford objects must be appended
circ.draw()

This error can be avoided by changing Clifford object to Instruction object by random_clifford(1).to_instruction().
That is the reason why the main version of Qiskit Experiments does not fail while the 0.4.0 fails. As @conradhaupt mentioned above, qiskit-community/qiskit-experiments#898 changed to always convert Clifford objects into Instruction objects before appending to a circuit, that results in avoiding the bug.

[conradhaupt]

@itoko: Ah! That makes sense.

[1ucian0]

This is related to #9159 as Clifford operations are not instructions.

[1ucian0]

Fixed by #8907