Avoiding Bad Qubits

The D-Wave 2000Q system has up to 2048 qubits, but the dashboard informs today that the "DW_2000Q_5" has 2030 qubits available. About the 18 bad-qubits, I wonder:

1) how to know what are (numbers) the bad-qubits?

2) how to avoid them when using "minorminer.find_embedding"?

3) what happens (in the QPU) if I manually specify a bad qubit in my embedding?

Thank you very much.



  • Hello,

    1) The bad qubits will not show up in the graph returned from the sampler.

    You can see this by calling the following code:

    sampler = DWaveSampler()
    sampler.nodelist # This call lists out all of the available qubits.

    You could then take check the difference between this list and a theoretical full list:

    set(range(16*16*8)) - set(sampler.nodelist)
    {577, 1730, 1858, 336, 1105, 1776, 691, 1012, 215, 1276}

    2) If you use minorminer.find_embedding, all of these qubits will be avoided automatically!

    3) The Ocean tools will not let you specify a bad qubit. You will get an error saying that the graph you submitted is not compatible with the sampler's graph:

    Traceback (most recent call last):
    File "<input>", line 1, in <module>
    File "C:\Users\djohnson\PycharmProjects\dwave\venv\lib\site-packages\dwave\system\samplers\dwave_sampler.py", line 276, in sample_ising
    raise BinaryQuadraticModelStructureError(msg)
    dimod.exceptions.BinaryQuadraticModelStructureError: Problem graph incompatible with solver.
    Comment actions Permalink
  • David,

    Basically is impossible for an embedding with one bad qubit to run.

    Thank you so much for all the details and examples.

    Comment actions Permalink
  • That's right but it should be possible to avoid this scenario with minorminer's find_embedding.

    Comment actions Permalink

Please sign in to leave a comment.

Didn't find what you were looking for?

New post