### Table 5: Statistical results obtained by this new version of the SES for the 13 test functions with 30 independent runs.

2004

"... In PAGE 10: ... The initial 0 was set to 0:001. 4 Results and Discussion The statistical results of the SMES for the first set of 13 test functions are summarized in Table5 and for the new set of 11 functions they are presented in Table 6. As described in Table 5, for the first 13 test problems the SMES was able to find the global optimum in seven (g01, g03, g04, g06, g08, g11 and g12) and it found solutions very close to the global optimum in the remaining six (g02, g05, g07, g09, g10, g13).... In PAGE 10: ...nd used in [9] is adopted. The initial 0 was set to 0:001. 4 Results and Discussion The statistical results of the SMES for the first set of 13 test functions are summarized in Table 5 and for the new set of 11 functions they are presented in Table 6. As described in Table5 , for the first 13 test problems the SMES was able to find the global optimum in seven (g01, g03, g04, g06, g08, g11 and g12) and it found solutions very close to the global optimum in the remaining six (g02, g05, g07, g09, g10, g13). These results show a competitive approach based on the current benchmark.... ..."

Cited by 1

### TABLE 1. Test problems. Function names in boldface indicates function was used in the comparison of the new and original C-GRASPs.

2006

Cited by 2

### TABLE 6. Average GAP value for each test function for the new C- GRASP. Boldface entries denote GAP values satisfying Inequality (2). Results are grouped by function dimensions.

2006

Cited by 2

### Table 1. Comparison of results between the previous version of our approach (denoted by CO- MOEA [4]), the new version (CO-MOEA2) and the NSGA-II [2], for all the test functions.

"... In PAGE 6: ... Test Function 3 Minimize f1(x1; x2) = x1; f2(x1; x2) = x2 subject to: g1(x1; x2) = x2 1 x2 2 + 1 + 0:1cos(16arctanx1 x2 ) 0 g2(x1; x2) = (x1 1 2)2 + (x2 1 2)2 1 2 0; 0:0 x1; x2 In this example, our approach used: popsizeinit = 100, popsizerec = 30 (40 gen). Table1 shows the values of the metrics for each of the MOEAs compared. 6 Discussion of Results As we can see in Table 1, in the first function the new version of our approach (CO- MOEA2) is clearly better than the previous version (CO-MOEA), with respect to all the metrics.... In PAGE 6: ... Table 1 shows the values of the metrics for each of the MOEAs compared. 6 Discussion of Results As we can see in Table1 , in the first function the new version of our approach (CO- MOEA2) is clearly better than the previous version (CO-MOEA), with respect to all the metrics. On the other hand, although the results of the NSGA-II are better on average than the results of CO-MOEA2, the SC metric indicates that the Pareto fronts obtained by both algorithms are on average almost of the same quality.... In PAGE 6: ... With respect to the SC metric, the NSGA-II obtained the best results, followed by CO-MOEA2 and CO-MOEA. In general, from Table1 and Figures 3 and 4, we can conclude that in the first two functions the new version of our approach has clearly improved the original version, and obtained very competitive results with respect to the NSGA-II. In the third func- tion, CO-MOEA2 has obtained the same quality on the results than the CO-MOEA and NSGA-II.... ..."

### Table 1. Test functions.

2004

"... In PAGE 3: ...igure 1. Obtaining a new proposal in DE. 3. Test Functions Used in Simulation Studies Five test functions (F1-F5) presented in Table1 have been rstly proposed by De Jong [11] to measure the performance of GAs. After Jong, they have been extensively used by GA researchers and other algorithm... ..."

Cited by 1

### Table 3: Primal dual vs primal options of the new interior point method. The table gives the number of function evaluations and percentage of full steps. A * indicates that the stopping test was not satisfied in 10,000 iterations.

"... In PAGE 22: ... A * indicates that the method did not meet the stopping test in 10,000 iterations. In Table3 we compare the primal version of NITRO using (3.1) and the primal-dual version using (3.... ..."

### Table 2. Simple Test Functions Function Constraints

1995

"... In PAGE 14: ... We compare this scaling rule for multiple mutations with a new individual step-size adaptation in Evolution Strategies [11]. As a test function the hyper- ellipsoid function Fellipsoid from Table2 with ?1 xi 1 is used. The average number of function evaluations lt;feval gt; for Evolution Strategies without accu- mulated information to nd the minimum with the speci ed accuracy is given in Table 4.... ..."

Cited by 8

### Table 2. The performance of various hybrid sampling strategies with a new reward function.

2004

"... In PAGE 17: ... We did some initial testing with this new reward function on the two environments in which the original reward function has unsatisfactory pre-computation time. The results, shown in Table2 , indicate that the new reward function indeed improves pre-computation time, if we compare the results with the corresponding ones for the last two environments in block 3a of Table 1. We are currently exploring efficient ways to implement this new reward function as well as other interesting reward functions.... ..."

Cited by 4

### Table 3: Primal dual vs primal options of the new interior point method. The table gives the number of function evaluations and percentage of full steps. A * indicates that the stopping test was not satis ed in 10,000 iterations.

1999

"... In PAGE 21: ... A * indicates that the method did not meet the stopping test in 10,000 iterations. In Table3 we compare the primal version of NITRO using (3.1) and the primal-dual version using (3.... ..."

Cited by 61

### Table 1. New instructions.

"... In PAGE 9: ...ontroller systematically. Additional hardware components, e.g. LFSR and MISR and suitable test instructions can be added to allow test methodology applied to the selected cores (see Table1 ). LFSR and MISR are not regular registers and new instructions for these two registers perform two jobs when these instructions are used.... In PAGE 19: ... A test evaluation framework for processors. Table1... ..."