Fig. 5
Fraction of instances where QAC outperforms the C strategy. In a, we plot the fraction of instances where QAC found higher success probability than the C strategy after removing any instances where both failed completely (see Supplementary Information Section B.1). In b, we plot the fraction of instances where QAC lowered the \(J\)-chaoticity measure when compared to the C strategy. QAC becomes better with increasing size. At large noise, QAC becomes the better strategy for all instances. We note that here and the other figures below, we have omitted the \(\eta =0\) results for \(L\ge 13\). The reason is a discontinuity between the DW2X and DW2000Q devices, which is discussed in detail in Supplementary Information Section C, and which is unrelated to the scaling analysis that forms the main focus of this work.
