### Table 4 Optimal sample size L* as a function of W under the assumption for L = 100

2005

"... In PAGE 10: ... Similarly as in the pre- vious section, we can determine which sample length is optimal under which conditions of the warmup length W. The results are summarized in Table4 . Again, we can make the same conclusions as in the previous section: the optimal sample length depends on the benchmark and increases with an increasing warmup length.... ..."

### Table 11.2 Comparison between the 2-Shewhart chart with optimal sample size and the 2-CUSUM algorithm.

### Table 2.1: Comparison of worths for three di erent approaches. In the table, n is optimal sample size, and ^ n is asymptotically optimal sample size. W is optimal worth, and ^ W is worth calculated at ^ n. The worth ~ W is calculated according to (2.25).

### Table 11.2 Comparison between the #1F 2 -Shewhart chart with optimal sample size and the #1F 2 -CUSUM algorithm.

### Table I. Sample program size characteristics (no optimization)

1992

Cited by 4

### Table 4. Sample Size Reduction (in %) of Optimal Interval over Asymmetric Interval

"... In PAGE 10: ...xed width. A larger value of h0 will result in a smaller total sample size Ni in (1). Let the width of the interval in both papers be xed at W = 1 unit. For given k, P and , the sample size reduction (in percentage) of the optimal interval over the asymmetric interval of Chen and Dudewicz (1976) and Chen (1977) is calculated by the ratio = 100(h0 hc)=hc (equivalent to (Nic Ni0)=Ni0), which is reported in Table4 . A positive and large ratio indicates that the amount of improvement in sample size reduction is signi cant.... In PAGE 10: ... A positive and large ratio indicates that the amount of improvement in sample size reduction is signi cant. We can see from Table4 that the sample size reduction of the optimal interval is signi cant and is uniformly better. For given k and , the sample size reduction increases as P increases; for example, when k is xed at 3 and = 1, the sample size reduction ranges from 78% at P = 0:80 to 83% at P = :99.... In PAGE 12: ...reduction ranges from 80% at = 1 reduced to 13% at = 120. For all calculations reported in Table4 , the sample size reduction ranges from 6% at k = 3, = 120, and P = :99 to 235% at k = 10, = 1, and P = :99. The general pattern of the sample size reduction for any other combinations of k, P and is similar to that of Table 4.... ..."

### Table 5.1: Numerical comparison of di erent schemes. W avg is the worth cal- culated at optimal sample size (n 1; n 2). ^ Wavg is the worth calculated at the asymptotic sample size (^ n1; ^ n2). ~ Wavg is the worth of the strategy in which one assigns N=4 patients to each of the two treatments in the rst stage and the remaining N=2 patients to the apparently better treatment in the second stage. For each case, the worth is calculated for N = 100; 200; 500 and 1000.

### Table I: Optimal number of samples s for sorting the [WR] integer benchmark on the Cray T3D, for a variety of processors and input sizes. Optimal Number of Samples s for Sorting on SP2 Number of Processors int./proc. 8 16 32 64 128

1996

Cited by 20

### Table 1: Stability tests for the optimization model. For every size of the scenario tree, 25 trees were generated, and the model was solved on each of them. The solutions were then evaluated on the benchmark tree to obtain the out-of-sample values. The table presents sample means and standard deviations of the optimal values, for the different sizes.

2003

"... In PAGE 10: ... This is repeated for several different sizes of the tree. Results of the test are presented in Table1 . We see that the scenario-generation method used gives a reasonable stability, both in-sample and out-of-sample.... ..."

Cited by 5