### Table 1: Approximation ratios achieved on our four maximization problems, our results appear in columns in which the SDP technique is mentioned.

1999

"... In PAGE 21: ... In general we have seen, for various values of k, that using semide nite programming one can obtain an approximation ratio strictly higher than that obtained using linear programming, which in turn improves the trivial ratio achieved by picking a subset U of size k at random. Table1 summarizes the approximation ratios achieved on our four maximization problems using the algorithmic techniques discussed in our work. We still do not fully understand how to exploit semide nite programs as approximation algorithms.... ..."

Cited by 27

### Table 10: Design that approximately robustly maximizes revenue.

in Automated mechanism design in infinite games of incomplete information: Framework and applications

2007

"... In PAGE 21: ... Since it is far from clear what the actual optimum for this problem or for its probably approximately robust equivalent is, we ran our automated framework to obtain an approximately optimal design. In Table10 we show the approxi-... In PAGE 21: ... Proposition 17. The mechanism in Table10 yields the value of 0.0066 for the robust objective.... ..."

Cited by 2

### Table 10: Design that approximately robustly maximizes revenue.

2007

"... In PAGE 21: ... Since it is far from clear what the actual optimum for this problem or for its probably approximately robust equivalent is, we ran our automated framework to obtain an approximately optimal design. In Table10 we show the approxi-... In PAGE 21: ... Proposition 17. The mechanism in Table10 yields the value of 0.0066 for the robust objective.... ..."

Cited by 1

### Table 4.4 shows the number of strips needed for a model. We can see that both SGI and STRIPE creates approximately three times less triangle strips than DTS. This is quite surprising because I have expected an algorithm that creates a low number of strips. This problem is caused by a big amount of flips during the triangulation process (6 flips per vertex on average).

2004

### Table 2: Results for the Complementary Algorithm

"... In PAGE 8: ...3 Performance of the Complementary Algorithm This section presents the performance results of the algorithm that generates the extra alignments. In the evaluation the recall and precision values and the F-measure values for the basic algorithm presented in Table 1, are compared to the values for recall and precision and the F-measure values of the complementary algorithm presented in Table2 . For both the closed-caption files and the autocue files the recall increased and the precision decreased.... ..."

### Table 2 Sets of 20 correlated problems (maximization)

2005

"... In PAGE 8: ... Again solutions were obtained rapidly and were near-optimal. The results in Table2 exhibit much of the same features as those in Table 1. Both sets of results significantly improved on those in Wilson [4], which used the same problem instances, where run times were longer and optimality levels poorer.... ..."

### Table 1 Approximation algorithms for the Vertex Cover Problem

2006

"... In PAGE 3: ... Several approximation algorithms have been proposed for the Unweighted and Weighted Versions of the Vertex Cover Problem. We list some of them in Table1 . We decided to implement these algorithms because of their simplicity and performance guarantees.... ..."

### Table 9: Performance Comparison for the Example strips Problem System CPU Time (sec.) Nodes Searched Solution Length

1994

"... In PAGE 42: ... First consider the results on the example problem described above. Table9 shows the CPU time, nodes searched and solution length. prodigy + alpine produces a small per- formance improvement over prodigy and generates shorter solutions.... ..."

Cited by 154

### Table 1 (continued)

"... In PAGE 5: ... We adopt the method for problem classiFFcation of Lageweg et al. [29] and present in Table1 the maximal easy problems (the most general cases of polynomially solvable problems) and the minimal hard problems (the most simple cases of NP problems). Other problems are cited below the table and are related to speciFFc problems described in the table.... ..."

### Table 1: Recovery of harmonic function

2007

"... In PAGE 14: ...Table1 concerns a more or less trivial case where the data come from a globally harmonic function f(x; y) = exp(x) cos(y) which is the real part of the entire complex function exp(z) = exp(x + iy). The real part of the power series of exp(z) yields a sequence of perfect approximations by harmonic polynomials on each domain whatsoever, and the recovery by collocation via har- monic polynomials even works on open arcs anywhere.... In PAGE 15: ... This is to be expected, because the solution has no nite singularities. The 25 marked source points in Figure 2 belong to the situation of the fth line of Table1 . The adaptive L2 algorithm picks these 25 source points with no connection to the domain corner, as is to be expected.... In PAGE 21: ...57e-002 1.62e+012 Table1 0: Sums of coe cients as functions of r... ..."