### Table 1: Description of the cases for which we evaluated the exact posterior marginals. Figure 3 shows the log-likelihood for the four tractable CPC cases. The gure also shows the variational lower and upper bounds. We calculated the variational bounds twice, with 14

1999

"... In PAGE 14: ... 5.1 Comparison to exact marginals Four of the CPC cases have 20 or fewer positive ndings (see Table1 ), and for these cases it is possible to calculate the exact values of the likelihood and the posterior marginals in a reasonable amount of time. We used Heckerman apos;s \Quickscore quot; algorithm (Hecker- man 1989)|an algorithm tailored to the QMR-DT architecture|to perform these exact... ..."

Cited by 12

### Table 2: Result Variations

"... In PAGE 15: ...Table 2: Result Variations Due to the randomized nature of the approach the results generated can vary from trial to trial. Table2 contains the minimum and maximum values for the % relief opportunities selected, TRACS II variables, and so forth found in the four (or ve in the case of the TRAM data) trials for the data used. Unfortunately, every time the ILP process is used with a di erent set of potential shifts the size of the branch and bound search will be di erent.... ..."

### Table 1: Variation of C

"... In PAGE 4: ... Therefore, the following relationship holds between the switching times and the equivalent coupling capacitance, C c;eq . The value of C c;eq in the #0Crst line of Table1 is chosen to be either 0 or 2C c , depending on whether the signals switch in the same direction, or in opposite directions. Note that it is possible for some of the aboveintervals to be empty when the lower bound and the upper bound of the interval coincide.... In PAGE 5: ...While the relationship shown in Table1 looks relatively straightforward, it is considerably complicated by the fact that d 1;min , d 1;max , d 2;min and d 2;max are dependent on the value of C c;eq , which is itself dependent on the values of d i;min and d i;max , i =1; 2. Therefore, an iterative approach is required.... In PAGE 5: ... We point out, though, that the methods described in this paper do not require equal rise and fall times and can be extended to unequal values using standard methods in timing analysis #28see, for example, #5B15#5D#29. On the surface, it would appear that none of the switching time intervals overlap, and an equivalent coupling capacitance of C c would prevail, based on Table1 . However, these switching intervals do not take the wire delayinto account, and hence we will now make that correction.... In PAGE 9: ... The neighbors of a wire j above correspond to adjacentvertices in the G s graph. The updates in lines 8, 10, 16 and 18 are performed using the scheme in Table1 , with the di#0Berence that the wire delays are calculated using the values of C c;eq based on the currentvalues of T start and T end for the nets. The update formul#1A are as follows: T end updates #0F If T end #28j#29 #3E T end #28i#29 #3E T start #28j#29, as in Figure 4#28a#29, then the worst case corresponds to an equivalent coupling capacitance of 2C c between wires i and j that is seen at T end #28i#29, resulting in the update shown by the dotted line.... ..."

### Table 5: Results for Gaussian variation sources.

2007

"... In PAGE 6: ... We also compare n2SSTA with our implementation of [2] (denoted as linSSTA) by assuming Gaussian variations and linear delay model for both. From Table5 , we see that in predicting = , n2SSTA matches Monte Carlo simulation well with about 5.5% error, while linSSTA has about 11% error.... In PAGE 6: ... This clearly shows that n2SSTA is not only more general, but also more accurate than linSSTA. Note that n2SSTA has a larger error for Gaussian variation sources in Table5 than for uniform or triangle variation sources in Table 4, and this is because n2SSTA needs bigger bounds (10) for Gaussian variations than for uniform or triangle variations. Interestingly, we nd that both approaches pre- dict the 95% yield point well.... ..."

Cited by 1

### Table 7.1: Parameters for the TVBM Runge-Kutta scheme parameters listed in Table 7.1 which were derived by Shu [48] to result in a time discretization that is TVD (Total Variation Diminishing) when combined with nite di erence schemes in one space dimension. For multiple space dimensions, the parameters listed in Table 7.1 result in a scheme which is TVBM (Total Variation Bounded in the Means) as de ned in the following lemma. Lemma 9 (Cockburn, Hou, and Shu [12]) Let a0 = inf x2 u0(x)

### Table 1: Timing analysis of STARI. d is the lower bound on the clock period (the upper bound is irrelevant for the correctness of STARI), skewt and skewr are the variations in skew to the transmitter and receiver, respectively. The delay through a C-element is set to [2; 3] in all cases. (i) is the maximum separation corresponding to equation (i). Correct operation of STARI is indicated with a checkmark in the OK column.

1995

"... In PAGE 29: ...Table1 shows the result for di erent values for the clock period ( ), and for di erent variations of clock skew to the transmitter (skewt) and to the receiver (skewr). The CPU time to verify the four conditions is less than a second on a SPARC 2.... ..."

Cited by 37

### Table 1. Parameters of variation for the Floating Weather Station product line (excerpts)

2005

"... In PAGE 7: ...hich the value is fixed (e.g., bound at compile time), and the default value. Table1 shows some parameters of variation for the FWS product line. (See (Padmanabhan, 2002) for a fuller treatment of the commonality analysis).... ..."

Cited by 2

### Table IV. Each entry lists a pair of events, the required minimum separation between these events, and the computed bounds on the separation during Phases I and II (lower and upper bounds shown are the minimum of the lower bounds and the maximum of the upper bounds obtained in the two phases). Table V shows the loop delay and cycle time of BWCXABBXD5 obtained with BCB1, A6BHB1 and A6BDBCB1 variations in delays. The BCB1 entries indicate the loop delay and cycle time variations solely due to data-dependent delay variations.

Cited by 1

### Table 1: Timing analysis of STARI. d is the lower bound on the clock period (the upper bound is irrele- vant for the correctness of STARI), skewt and skewr are the variations in skew to the transmitter and re- ceiver, respectively. The delay through a C-element is set to [2;3] in all cases. (i) is the maximum separa- tion corresponding to equation (i). Correct operation of STARI is indicatedwith a checkmarkin the OK column.

1993

"... In PAGE 5: ... Thus, by applying our algorithm four times we can verify the correct operation of the STARI protocol for all possible delay variations in the speci ed ranges. Table1 shows the result for di erent values for the clock period ( ), and for di erent variations of clock skew to the transmitter (skewt) and to the receiver (skewr). The CPU time to verify the four conditions is less than a second on a SPARC 2.... ..."

Cited by 11

### Table 2.1: General attributes and settings.

2007

"... In PAGE 44: ...Table2 deal more with the bounds of the variation of the hatches. Examples include the minimum and maximum number of rows and columns of hatches or the extents of how far a hatch can be rotated from its normal direction.... ..."