I want to know exactly how to estimate the Avg CNOT Error rate for IBMQ Backends?

For instance, I tried to estimate the Avg CNOT Error rate for Belem Backend; I randomly prepared the 00,01,10,11 states, applied a CNOT between those qubits, and got the results, but I couldn't get a similar value for the interaction Error for the 0 qubit and the 1 qubit, which is written on an excel file (ibmq_belem_calibrations).

If I can simulate the Error rate estimation with a quantum circuit, then how can I do this?


1 Answer 1


From reading your post I assume that you calculate the CNOT error by:

  • preparing one of the four states from {00,01,10,11}
  • Applying a CNOT gate
  • Measuring the final state
  • The probalitiy of measuring {00,01,11,10}, respectively, is then what you (?) assume to be the cnot error

If you follow this procedure then your 'CNOT error rate' will also be affected by errors in preparation, measurent and decoherence.

If I am not mistaken then IBMQ uses a method known as randomized benchmarking to determine the gate errors. Perhaps, this link can help you in calculating the CNOT error rate: Randomized Benchmarking with Qiskit

If you follow this procedure you should find a similar error rate. Take note that calculating errors with this method is a stochastic process and that error rate vary over time.

  • $\begingroup$ Thank you very much for your answer, and from your information I should check on the site where you recommend to read. I will definitely see how things are going. I appreciate your information. ^^ $\endgroup$ Jul 3, 2021 at 5:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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