"... In PAGE 6: ... This is due to fault masking after removal of the inert subsequences, which is assumed not to occur in our implementation of criterion 3. The compaction results for GATEST test sets shown in Table5 display trends similar to the HITEC com- paction results. However, GATEST was targeted at gen- erating compact test sets, and because GATEST test sets are much more compact than HITEC test sets to begin with, less compaction is obtained.... ..."

"... In PAGE 7: ... First, the fault coverages obtained are di erent, and second, the execution times needed are also di erent. Table7 compares our results with the static compaction technique proposed in [3] and the dynamic compaction proposed in [11] for the HITEC test sets. Fault coverage, test set size, and time for com- paction are shown for each technique after completion of the compaction.... ..."

"... In PAGE 6: ... Here, compaction was measured by performing fault simulation using the test set that achieved the higher fault coverage and noting the num- ber of vectors required to achieve the lower fault cover- age. For circuits having differences in fault coverage for the two techniques, the number of vectors needed from each of the test sets to obtain the lower fault coverage is shown in Table2 . Since the test sequences evolved by Squeeze detect more faults than those evolved by GA- COMPACT, fewer faults had to be targeted by the test... ..."