### Table 7 Computation results for 48 h schedule of Example 1

"... In PAGE 8: ...value and resolving the original scheduling problem of 48 h. The result obtained by this approach is shown in Table7 and compared with the solution of the original problem and the solution of the overall problem by applying cyclic scheduling mode which repeatedly executes the same operation schedule for a smaller time horizon. As illustrated in Table 7, the proposed approach generates an improved objective function value reducing the computational time in more than one order of magnitude.... In PAGE 8: ... The result obtained by this approach is shown in Table 7 and compared with the solution of the original problem and the solution of the overall problem by applying cyclic scheduling mode which repeatedly executes the same operation schedule for a smaller time horizon. As illustrated in Table7 , the proposed approach generates an improved objective function value reducing the computational time in more than one order of magnitude. The same approach was applied to Example 2.... ..."

### Table 9 Computation results for 48 h schedule of Example 2

"... In PAGE 8: ... The objectiveisto maximize the profit with the given time horizon. The computational results of the proposed approach, origi- nal scheduling problem and cyclic mode approach are listed in Table9 , pointing to the fact that the proposed approach performs very well since a very good solution in terms of objective value is obtained in reasonably short time. 4.... ..."

### Table 1. Performance results of both the RTS and the MF algorithms.

### Table 13: Sampling results for the 2-regime mixture model in (45)

2004

"... In PAGE 27: ...o differ no more than 0.10, then convergence has not been reached after 25000000 drawings. However, if we draw 2000000 points from the t1 distribution around the maximum likelihood estimator, and iteratively update the mean and covariance matrix, then we reach convergence after 10 iterations. Table13 shows the results. The methods yield approximately the same results, where the AdMit procedures require less drawings and time.... ..."

### Table 6: Neural network based sampling results for the Bayesian IV regression

2004

### Table 9: Sampling results for the Bayesian analysis of a VECM

2004

"... In PAGE 21: ...l2 are less than 0.005, 0.005 and 0.05, respectively. Table9 shows the results. Note that IS and MH with Student t candidate density again require very large amounts of drawings to reach convergence.... ..."

### Table 11: Neural network based sampling results for the 2-regime mixture model (43)

2004

"... In PAGE 23: ...05. Table11 and 12 show the results. Even after 25 million drawings IS and MH with the normal candidate distribution yield completely different results than the other algorithms.... ..."

### Table 1 shows the particular parameter settings for our evaluation system when all nodes are equipped with an A203 carrier board. Name Value Meaning

1999

"... In PAGE 9: ... Table1... In PAGE 11: ... (3). This step ultimately allows us to neglect the rare excessive transmission delays apparent in Figure 7 and choose quot;ps = [?3 s; 3 s] in Table1 instead: Since CV tolerates up to f faulty input arguments, a larger f (i.e.... In PAGE 11: ... Since we established in [Sch97b] and [SS99] that the result of a suitable convergence function is contained in the intersection of its m ? 2f-largest non-faulty input intervals, it follows that Ip s yields node s apos;s interval of accuracies s immediately after resynchro- nization. Plugging in the simpli ed parameters from Table1 and the obvious bounds Ts; Tp , we obtain s = 0 + 2u + G + 1s + u + G + quot; + (3 + ?) = 0 + 3u + 2G + quot; + (3 + ? + 1s) : Moreover, the interval of accuracies immediately before the next resynchronization is bounded by 0 = + P + u + G for G = [0; G], so that the worst case precision of any two non-faulty S-nodes in the system evaluates to = j 0j. Plugging in our particular parameter values nally gives the worst case accuracy and precision shown in Table 2.... ..."

Cited by 11

### Table 1 shows the particular parameter settings for our evaluation system when all nodes are equipped with an A203 carrier board. Name Value Meaning

1999

"... In PAGE 9: ... Table1... In PAGE 11: ... (3). This step ultimately allows us to neglect the rare excessive transmission delays apparent in Figure 7 and choose quot;ps = [?3 s; 3 s] in Table1 instead: Since CV tolerates up to f faulty input arguments, a larger f (i.e.... In PAGE 11: ... Since we established in [Sch97b] and [SS99] that the result of a suitable convergence function is contained in the intersection of its m ? 2f-largest non-faulty input intervals, it follows that Ip s yields node s apos;s interval of accuracies s immediately after resynchro- nization. Plugging in the simpli ed parameters from Table1 and the obvious bounds Ts; Tp , we obtain s = 0 + 2u + G + 1s + u + G + quot; + (3 + ?) = 0 + 3u + 2G + quot; + (3 + ? + 1s) : Moreover, the interval of accuracies immediately before the next resynchronization is bounded by 0 = + P + u + G for G = [0; G], so that the worst case precision of any two non-faulty S-nodes in the system evaluates to = j 0j. Plugging in our particular parameter values nally gives the worst case accuracy and precision shown in Table 2.... ..."

Cited by 11

### Table 3: An experimental result of ten runs on data set 2 (N=100, K=20)

2003

"... In PAGE 5: ...47GHz speed and 240 MB RAM. The computational results are shown in Table 2 and Table3 and are obtained from 10 runs for each data set. Constructive Heuristic Improvement Heuristic Test CT (unit time) P(seconds) CT (unit time) P(seconds) Improvement (%) 1 52.... In PAGE 6: ...Table3 show that our greedy constructive heuristic can generate an initial solution in a short time of about 0.070 seconds for data set 1 and 0.... ..."

Cited by 7