### Table 1: Quantity of text retained by the agent

"... In PAGE 3: ... When the agent was directed to the sites described above, their contents were analysed and a number of documents retained. The top three sites (in terms of quantity of text retained) are shown in Table1 . The quantity of text retained from the remaining sites was either negligible or zero.... ..."

### Table 1: Convergence of E[SK;M] to its limit E[SK]. The quantity E[SK] is obtained by simulation with a 95% con dence interval. In the rst three cases listed in Table 1 the con dence interval has been obtained from 20 runs of 106 arrivals. In the last three cases 40 runs of 107 arrivals were used. The computational e ort to obtain the quantities E[SK;M] is negligible compared to the e ort to obtain E[SK]. We see that already for small values of M, E[SK;M] is a good approximation for E[SK].

"... In PAGE 17: ... Finally, we show the trade-o between extra delay on one hand and reduction of burstiness of the output on the other hand, when the maximum amount of credit decreases. In Table1 , the mean sojourn time E[SK] in the model with nite amount of credit K is compared with the mean sojourn time in the model with nitely many portions of credit for several values of the parameters. We denote the mean sojourn time in case credit is split up into at most M exponential phases with mean K=M by E[SK;M].... ..."

### Table 1: Object quantities Quantity Quantity space Quantity Quantity space

### Table 3 For different design densities we compare the medians of the same quantities as in Table 2

2005

"... In PAGE 22: ...ver f2 because of the high peak in rtrue at (0.75, 0.75). This is reflected by the optimal penalty Ropt: compared with the uniform design, f1 needs a larger and f2 a smaller penalty; see Table3 (e). Similarly, the ideal relative gain due to penalizing is larger for f1 and smaller for f2, item (a).... In PAGE 23: ... x = ISE(AIC)]. When doubling n to 400, parameter selection is improved; see Table3 , col- umn 400, item (b). Because of the smaller Ropt, the gain due to penalizing is smaller but still not negligible.... ..."

### Table 2b. Optimal uniform partition sizes for the PLH. For these partitions are pre- sented the quantities analogous to those in Table 1.

"... In PAGE 15: ... In Tables 2a { 2c one can observe some pleasant and some unpleasant facts. A pleasant fact is that for all sample sizes 50 n 10 000, the GPLH with the optimally adapted partitions achieved the absolute minima of the \means quot; : = MIAE apos;s presented in these tables (the deviations from this rule are due to FP for 100 n 500 and for the partitions optimally adapted to the PLH in Table2 b; quantitatively they are of the negligible order of 10?3). Another pleasant fact partly observable in these tables is that the partitions optimal for the GPLH apos;s in sense of the theory in Section 4, really minimize the expected values of IAE apos;s (to this end compare the \means quot; of GPLH in Table 2a with those in Table 2b for n = 100 and 500 n 10 000, and with those in Table 2c for all 50 n 10 000).... In PAGE 15: ... A pleasant fact is that for all sample sizes 50 n 10 000, the GPLH with the optimally adapted partitions achieved the absolute minima of the \means quot; : = MIAE apos;s presented in these tables (the deviations from this rule are due to FP for 100 n 500 and for the partitions optimally adapted to the PLH in Table 2b; quantitatively they are of the negligible order of 10?3). Another pleasant fact partly observable in these tables is that the partitions optimal for the GPLH apos;s in sense of the theory in Section 4, really minimize the expected values of IAE apos;s (to this end compare the \means quot; of GPLH in Table2 a with those in Table 2b for n = 100 and 500 n 10 000, and with those in Table 2c for all 50 n 10 000). An unpleasant fact which is clearly observable in the tables is that the partitions optimal for PLH and FP in sense of the theory in Beirlant et al (1999) do not minimize the expected IAE apos;s: the \means quot; of PLH in Table 2a are uniformly less than (or equal to) those in Table 2b and the \means quot; of FP in Table 2b are considerably less than those in Table 2c.... In PAGE 15: ... A pleasant fact is that for all sample sizes 50 n 10 000, the GPLH with the optimally adapted partitions achieved the absolute minima of the \means quot; : = MIAE apos;s presented in these tables (the deviations from this rule are due to FP for 100 n 500 and for the partitions optimally adapted to the PLH in Table 2b; quantitatively they are of the negligible order of 10?3). Another pleasant fact partly observable in these tables is that the partitions optimal for the GPLH apos;s in sense of the theory in Section 4, really minimize the expected values of IAE apos;s (to this end compare the \means quot; of GPLH in Table 2a with those in Table2 b for n = 100 and 500 n 10 000, and with those in Table 2c for all 50 n 10 000). An unpleasant fact which is clearly observable in the tables is that the partitions optimal for PLH and FP in sense of the theory in Beirlant et al (1999) do not minimize the expected IAE apos;s: the \means quot; of PLH in Table 2a are uniformly less than (or equal to) those in Table 2b and the \means quot; of FP in Table 2b are considerably less than those in Table 2c.... In PAGE 15: ... Another pleasant fact partly observable in these tables is that the partitions optimal for the GPLH apos;s in sense of the theory in Section 4, really minimize the expected values of IAE apos;s (to this end compare the \means quot; of GPLH in Table 2a with those in Table 2b for n = 100 and 500 n 10 000, and with those in Table 2c for all 50 n 10 000). An unpleasant fact which is clearly observable in the tables is that the partitions optimal for PLH and FP in sense of the theory in Beirlant et al (1999) do not minimize the expected IAE apos;s: the \means quot; of PLH in Table2 a are uniformly less than (or equal to) those in Table 2b and the \means quot; of FP in Table 2b are considerably less than those in Table 2c. In other words, the partition sizes mn in Table 2b (for PLH) are overestimated slightly and in Table 2c (for FP) quite considerably.... In PAGE 15: ... Another pleasant fact partly observable in these tables is that the partitions optimal for the GPLH apos;s in sense of the theory in Section 4, really minimize the expected values of IAE apos;s (to this end compare the \means quot; of GPLH in Table 2a with those in Table 2b for n = 100 and 500 n 10 000, and with those in Table 2c for all 50 n 10 000). An unpleasant fact which is clearly observable in the tables is that the partitions optimal for PLH and FP in sense of the theory in Beirlant et al (1999) do not minimize the expected IAE apos;s: the \means quot; of PLH in Table 2a are uniformly less than (or equal to) those in Table2 b and the \means quot; of FP in Table 2b are considerably less than those in Table 2c. In other words, the partition sizes mn in Table 2b (for PLH) are overestimated slightly and in Table 2c (for FP) quite considerably.... In PAGE 15: ... Another pleasant fact partly observable in these tables is that the partitions optimal for the GPLH apos;s in sense of the theory in Section 4, really minimize the expected values of IAE apos;s (to this end compare the \means quot; of GPLH in Table 2a with those in Table 2b for n = 100 and 500 n 10 000, and with those in Table 2c for all 50 n 10 000). An unpleasant fact which is clearly observable in the tables is that the partitions optimal for PLH and FP in sense of the theory in Beirlant et al (1999) do not minimize the expected IAE apos;s: the \means quot; of PLH in Table 2a are uniformly less than (or equal to) those in Table 2b and the \means quot; of FP in Table 2b are considerably less than those in Table2 c. In other words, the partition sizes mn in Table 2b (for PLH) are overestimated slightly and in Table 2c (for FP) quite considerably.... In PAGE 15: ... An unpleasant fact which is clearly observable in the tables is that the partitions optimal for PLH and FP in sense of the theory in Beirlant et al (1999) do not minimize the expected IAE apos;s: the \means quot; of PLH in Table 2a are uniformly less than (or equal to) those in Table 2b and the \means quot; of FP in Table 2b are considerably less than those in Table 2c. In other words, the partition sizes mn in Table 2b (for PLH) are overestimated slightly and in Table2 c (for FP) quite considerably. The partition sizes of Table 2a (for GPLH) practically optimize all three estimators simultaneously.... In PAGE 15: ... In other words, the partition sizes mn in Table 2b (for PLH) are overestimated slightly and in Table 2c (for FP) quite considerably. The partition sizes of Table2 a (for GPLH) practically optimize all three estimators simultaneously. GPLH PLH FP n mn(GPLH) mean stdev % mean stdev % mean stdev 50 2 .... In PAGE 15: ...028 .006 Table2 a. Optimal uniform partition sizes for the GPLH under consideration.... In PAGE 16: ... Table2 c. Optimal uniform partitions for the FP.... ..."

### Table 1: Estimated Quantities Quantity Source Description

### Table 2: Physical Quantities

1999

"... In PAGE 33: ... It is desired to save heavy oil by supplying by-product gas to boilers at a stationary rate. The physical quantities describing the model and the numerical values of the parameters are reported in Table2 and 3 respectively. Table 2: Physical Quantities... ..."

Cited by 97

### Table 2: Physical Quantities

1999

"... In PAGE 33: ... It is desired to save heavy oil by supplying by-product gas to boilers at a stationary rate. The physical quantities describing the model and the numerical values of the parameters are reported in Table2 and 3 respectively. Table 2: Physical Quantities... ..."

Cited by 97

### Table 2 Physical quantities

1999

Cited by 97

### Table 4. NEMD data. Summary of the scaling behaviors of relaxation times (upper half) and of rheological and structural quantities (lower half) as defined and discussed in the text.

"... In PAGE 16: ... Values for are summarized in Figure 9 and Table 3. When plotting the curves, the relaxation times G were used (see Table4 for details). Previously we reported A;B;C = eq 0:58 in equilibrium.... In PAGE 18: ...g. in [45], does not apply (see Table4 ). The normal stresses produced by the pure solvent are negligibly small compared to those measured for the model polymer solution.... In PAGE 18: ... These deviations were related to the flow-induced changes of the distribution of bond lengths, which is particularly affected by the FENE potential. (iv) Orientational anisotropy: The flow-alignment angles A and G, which are experimen- tally accessible via flow birefringence and light scattering measurements [2, 4, 5, 44, 45], have been shown to scale with chain size N and shear rate (see Table4 ), in agreement with exper- imental findings. It has been demonstrated that, in contrast to the commonly referenced the- oretical predictions cot(2 ) / , the decrease of alignment angle ( ) with increasing shear rate is more pronounced at low rates and weaker at high rates.... In PAGE 19: ...g., / N1:6 for X = int (see Table4 for details). In contrast to theoretical predictions, these power laws depend on the chosen quantity X, reflecting the underlying influ- ence of relaxation processes which act on different length scales.... In PAGE 27: ... NEMD data for the shear-induced intrinsic shear viscosity int, the radius of gyration rG, the flow-alignment angle of the radius of gyration G and of the polymer segments A, the orientational resistance parameter mG and the reduced angular velocity of the polymer coil ?!z= . The applied shear rate (in reduced LJ units) is related to the dimensionless shear parameter via D = D (see Table4... ..."