Fig. 4. Performance of the six different search approaches on the precalculated benchmark dataset of 30 000 molecules. The solid-coloured lines show the mean Fcomb over 25 runs with different initial populations and the coloured shaded area shows the variance of the Fcomb over those different runs; (a) maximum Fcomb found for an oligomer up to the current iteration. The histogram on the right shows the distribution of the oligomers in the benchmark dataset; (b) mean Fcomb of the oligomers found up to the current iteration; (c) number of oligomers in the top 1% found up to the current iteration (top 1% is 300 molecules); (d) discovery rate of the top 1% oligomers in the dataset, calculated as the (number of top molecules found)/(number of iterations).
