Do the readout errors on the publicly available IBM quantum computers have any dependency on the state being measured? That is, if we are measuring a qubit in the state

$\cos({\theta/2}) |0> + e^{i\phi} \sin({\theta/2}) |1>$ ,

does the error rate have any $\theta$ or $\phi$ dependence? If so, what does the dependence look like?

Thanks in advance for any replies.


1 Answer 1


the answer is yes - in general $p(m=0|s=1)> p(m=1|s=0)$, where $s$ is the state at the beginning of the measurement, and $m$ is the measurement outcome.

This is because the qubit has a probability to decay from 1 to zero during the measurement which is much higher than the probability of a false excitation from 0 to 1, by the Boltzmann factor $\exp({\hbar\omega_{01}/k_BT})$. So the dependence is on $\theta$, and it scales approximately as $\sin^2\theta/2$.

This is disucssed explicitly, for example, in fig. 2 here.

  • $\begingroup$ Thanks for your answer! I don't completely understand yet where the $\sin^2 \theta/2 scaling comes from, could you elaborate a bit more on that? $\endgroup$
    – user206444
    Mar 21 at 11:19
  • 1
    $\begingroup$ sure, there's nothing smart about it - all I'm saying is that the error rate is $p_e = p(0|1)p(1) + p(1|0)p(0)$, and since the $p(0|1)$ and $p(1|0)$ are system parameters, and $p(0|1)$ is dominant, $p_e \simeq p(0|1)p(1) = p(0|1)\sin^2\theta/2$. $\endgroup$
    – Lior
    Mar 21 at 13:04
  • $\begingroup$ Ah, I see thank you. $\endgroup$
    – user206444
    Mar 21 at 13:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.