Consider the circuit below.

This is almost the same as the standard protocol to perform a non-local $CNOT_{0,3}$. The only difference is that I decomposed the upper local $CNOT_{0,1}$ into one Ising gate and some local rotational gates.

When I attempt a simulated process tomography over qubits $q_0,q_3$ and evaluate the fidelity w.r.t. the ideal $CNOT$ I get a fidelity $\approx$85%, with 100k shots.

Why does this happen? Does such a transpiling mess with the logic behind the non-local $CNOT$?



Your Answer

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