I'm trying to reproduce some threshold results using 7-qubit Steane code in those legacy papers. I managed to obtain the graph for circuit noise and single qubit gate e.g. the pseudocode for computing the threshold of transversal H in each run would be:
1. Apply transversal H gate
2. Error correction using Steane fault-tolerant method, i.e. two encoded ancillas
3. Apply a perfect decoder
4. If there is a logical error, failure += 1
I tried to do similar thing for transversal CNOT like this
1. Apply transversal CNOT to 14 qubits
2. Error correction on each block of 7 qubits
3. Apply two perfect decoders to both logical qubits
4. If any of the two qubits has logical error, failure += 1
But this procedure doesn't give me a quadratic curve, instead it's giving a straight line. Could anyone point out what I've done wrong? Thanks.
My graph for CNOT looks like the following, which is clearly wrong (no threshold found!)
