"... In PAGE 11: ... The terms letter, legal, and ledger mean very little to other countries. Table6 demonstrates the different paper sizes in other countries.... ..."

"... In PAGE 43: ... Experimental comparisons among different metaheuristics At the mo- ment, most of the papers in the SCOP literature focus on one single metaheuris- tic, which is compared either to variants of the same metaheuristic, or to simple heuristics such as random search, or to exact methods when these are available. Only a very small number of papers perform comparisons among different meta- heuristics, as reported in Table5 . Thus, it is still impossible to give guidelines on which metaheuristic is better in which situation.... ..."

"... In PAGE 8: ... In other words, the sketch expressed optimization ideas which the compiler was unable to discover on its own. The results of the performance evaluation of the synthesizer are summarized in Table1 . The table lists all the benchmarks... In PAGE 9: ... The sketch for the case where N is even is very simple; it has two loop-nests with unspecified bounds, each with four assignments of the form out[2*i ??,2*j ??,2*k ??] = F(2*i ??,2*j ??,2*k ??, in); The sketch describes the high-level idea that we compute first all the red cells four at a time, and then all the black cells, also four at a time. The sketches rb2d1 and rb3d1 from Table1 correspond to the 2-D and 3-D instances of this sketch respectively. One can see that both sketches resolved quite fast despite having a large number of holes.... ..."

"... In PAGE 8: ... In other words, the sketch expressed optimization ideas which the compiler was unable to discover on its own. The results of the performance evaluation of the synthesizer are summarized in Table1... In PAGE 9: ... The sketch for the case where N is even is very simple; it has two loop-nests with unspecified bounds, each with four assignments of the form out[2*i-??,2*j-??,2*k-??] = F(2*i-??,2*j-??,2*k-??, in); The sketch describes the high-level idea that we compute first all the red cells four at a time, and then all the black cells, also four at a time. The sketches rb2d1 and rb3d1 from Table1 correspond to the 2D and 3D instances of this sketch respectively. One can see that both sketches resolved quite fast despite having a large number of holes.... ..."

"... In PAGE 8: ... In other words, the sketch expressed optimization ideas which the compiler was unable to discover on its own. The results of the performance evaluation of the synthesizer are summarized in Table1 . The table lists all the benchmarks... In PAGE 9: ... The sketch for the case where N is even is very simple; it has two loop-nests with unspecified bounds, each with four assignments of the form out[2*i ??,2*j ??,2*k ??] = F(2*i ??,2*j ??,2*k ??, in); The sketch describes the high-level idea that we compute first all the red cells four at a time, and then all the black cells, also four at a time. The sketches rb2d1 and rb3d1 from Table1 correspond to the 2-D and 3-D instances of this sketch respectively. One can see that both sketches resolved quite fast despite having a large number of holes.... ..."