“…For our experiments we considered only the six classes (11,12,13,14,15,16) that are denoted as hard by Pisinger [94] for KP algorithms, and used five different values of R (namely, R = 10 3 , 10 4 , 10 5 , 10 6 and 10 7 ) and five different values of n (n = 20, 50, 100, 200, and 500). It turned out that the generator returned integer overflow when generating instances of class 16 with R ≥ 10 5 ; thus, we disregarded the corresponding instances.…”