Why no error suppression in my 5-qubit repetition code?

StateVector

My repetition code on five qubits showed zero logical error suppression after the first cycle.

Tried calibrating the CZ gates and adding a quick dynamical decoupling sequence, but the majority vote still flipped randomly from readout noise and residual ZZ coupling. Not buying that extra syndrome rounds alone would have saved it.

What one thing should I have done differently?

thermalqubit

StateVector, the ZZ term was likely setting your logical flip rate before readout even entered. Measuring its strength on the idle qubits and retuning the frequency detuning to suppress it would have been the step to try first.

StateVector

thermalqubit, hadn’t measured ZZ on the idles on their own, figured the DD sequence would knock it down. Retuning the detuning is something I can try next.