Figure 4.

(a) Testing the sampling operator using a large multiset made of 1000 unique elements with 1000 different copy numbers. Images describe linear and log-scale representation of the confidence interval of the sampling operator. Solutions beyond this interval were not observed in 5000 trials. Dotted line represents an overestimate of the 99.9% confidence interval (for details, see Figure S4). Most probable outcomes of the S a operator have either zero or one unique sequence beyond this interval. This line is used in subsequent sections (Figures 5 and 6). We note that distributions of the copy numbers have well-defined shape; according to central limit theorem, it is a normal distribution. With enough replicas, it should be possible to extrapolate the center of this distribution, define the solutions explicitly, and bypass the stochastic nature of the S a operator.