### Table 2: Number of controller refreshments in nonlinear simulations.

1997

"... In PAGE 11: ...: dotted every 50 deg.: dashdot; no refreshment: dashed The simulations corresponding to di erent refreshment strategies, Table2 , are compared in Figure 3 with the fully-continuous simulations, that are obtained using the original continuous gain-scheduled con- troller. It can be seen that when the discrete controller is refreshed at each sample, the time responses are... In PAGE 12: ...using the trapezoidal method with a sampling frequency of 50 Hz. When the discrete controller is refreshed every 30 degrees of observed deviation the total amount of ops is roughly divided by 120 ( Table2 ) without signi cant performance degradation. With fewer refreshments (every 50 degrees), alterations of the per- formance are more visible in the transient and steady-state zones of the responses.... ..."

Cited by 2

### Table 2: Number of controller refreshments in nonlinear simulations.

1997

"... In PAGE 11: ...: dotted every 50 deg.: dashdot; no refreshment: dashed The simulations corresponding to di erent refreshment strategies, Table2 , are compared in Figure 3 with the fully-continuous simulations, that are obtained using the original continuous gain-scheduled con- troller. It can be seen that when the discrete controller is refreshed at each sample, the time responses are... In PAGE 12: ...using the trapezoidal method with a sampling frequency of 50 Hz. When the discrete controller is refreshed every 30 degrees of observed deviation the total amount of ops is roughly divided by 120 ( Table2 ) without signi cant performance degradation. With fewer refreshments (every 50 degrees), alterations of the per- formance are more visible in the transient and steady-state zones of the responses.... ..."

Cited by 2

### Table 3 shows the relationship between the grade of consistency, refreshing tim- ing and refreshing modes. For example, for o -line, and continuous refreshing (re- fresh after every update), the highest level of consistency, which can be achieved, is Completeness; and for o -line, and periodical refreshing, the highest reachable level of consistency is Strong Consistency. On the other hand, for on-line re- freshing, the level of consistency can not be determined, since it still depends on other factors, such as the activeness of the sources, self-maintainability of the warehouse data, etc. Table 3 is not complete. It can be expanded to include more factors, and can then serve as a guideline for determining what level of consis- tency can be achieved by using a certain refreshing mode, strategy or technique.

1997

"... In PAGE 12: ... Table3 . Levels of consistency achieved by di erent refreshing strategies Data purging Another maintenance issue of the DW is the purging of the old data, which is rarely mentioned in the literature.... ..."

Cited by 18

### Table 1 shows performance results of the HEF strategy with varying poll periods. We expected that the staleness would increase as a function of the poll period. To the contrary, no visible difference was found. This is some- what surprising to us and warrants further investigation. The final results are about the effect of the refreshment granularity sizes on performance of the HEF strategy (Ta- ble 2). We see that the staleness values marginally de-

"... In PAGE 9: ...12090 0.12096 Table1 : Time averaged staleness values (STN) for the HEF strategy with varying poll periods (PP). The 0/1 stal- eness is used.... ..."

### Table 6.8. Data Packets with Medium Mobility for Refresh2, Refresh2N and Refresh3 Refresh2 Refresh2N Refresh3

### Table 4. Required Refreshment Timescales

"... In PAGE 10: ... Balancing the addition of M (by accretion in the CNM, with timescale a;c) against the removal of M (during grain destruction, with timescale d, and star formation, with timescale in), we nd (M) = fc cxcos a;c ! M mC mgra 1 in + 1 d ?1 ; (24) where mgra is the mass of the entire carbonaceous grain population, per H atom. In Table4 we display max re (10 A)M=( ? ), and (M) for several elements, adopting mgra = 5:3 10?27g=H. The values of a;c to be used in equation (24) are determined from equations (18) and (19) by demanding that d 6 108 yr, and are denoted req a;c in Table 4; the values of (48=A)1=2s for use in equation (23) follow from these required values of a;c, assuming distribution (ii).... In PAGE 10: ... In Table 4 we display max re (10 A)M=( ? ), and (M) for several elements, adopting mgra = 5:3 10?27g=H. The values of a;c to be used in equation (24) are determined from equations (18) and (19) by demanding that d 6 108 yr, and are denoted req a;c in Table4 ; the values of (48=A)1=2s for use in equation (23) follow from these required values of a;c, assuming distribution (ii). Some metals might accrete onto only the silicates, resulting in the growth of these grains, and not contribute to the need to refresh the very small carbonaceous grains.... In PAGE 11: ... Tielens (1998) has recently inferred that the timescale for the return of Si from grains to gas is about an order of magnitude less than that for Fe, since Si is much less depleted in the warm phase. In computing our Table4 , we took the same d for each species, resulting in di erent transfer timescales, as exempli ed by mw, the time for transfering gas from molecular clouds to the warm phase. Of course, mw should be the same for each species.... ..."

### Table 1: Standard and extended refresh rate

"... In PAGE 11: ... If the result is 125 sec, it is called Extended Refresh device. Table1 shows the refresh time for di erent DRAM technologies. 2.... ..."

Cited by 1

### Table 2: Key-refreshment interval

"... In PAGE 18: ... Thus, one can set the key-refreshment interval to an appropriate value while allowing a small number of collisions. Table2 shows the required key-refreshment intervals for various values of ncollision. Clearly, the generalized criterion makes a security trade-off by the choice of ncollision.... ..."

### Table 1: Key-refreshment interval

"... In PAGE 16: ...igure 11 plots the probability of IV collision vs. the number of data frames. Then the criterion is to find the largest k (the number of data frames) such that Pa55 na59 ka56 a1 pcollision a0 We can calculate required key-refreshment intervals for various IV collision rates. Table1 summarizes the result, where the frame size and the transmission rate are set to 1,500 [Bytes] and 11 [Mbps], respectively. Table 1: Key-refreshment interval... ..."