Fig. 6: Effects of flag gadgets.
From: Efficient fault-tolerant implementations of non-Clifford gates with reconfigurable atom arrays
The rate (a) and output error (b) with and without flag gadgets, with data obtained by 200 rounds Monte Carlo simulation for each point. The input ancillae are accurate. The rate is suppressed linearly with flag since several single CNOT error now contributes linearly to the − 1 flag measurement instead of the output error. The output error shows a behaviour of 3.5p and 2.5p with or without flag gadget, respectively.
