### Table 1: Performance improvement with non-zero clock skew

"... In PAGE 5: ... The set-up and hold times are set to zero. The performance results are illustrated in Table1 . The number of registers and gates within the circuit including the I/O registers... In PAGE 6: ...The results shown in Table1 clearly demonstrate reductions of the minimum clock period when intentional clock skew is exploited. The amount of reduction is dependent on the characteristics of each circuit, particularly the differences in propagation delay between each local data path.... ..."

### Table 1: Non-zero coe cients and scaling factors of the -rotations W

1997

"... In PAGE 3: ....2.1. Example: N = 128, M =8 Figure 7 shows the frequency response of a length-128 pro- totype for an 8-channel cosine-modulated lter bank de- signed with the QCLS-algorithm [3] and of the prototype allowing an e cient realization. The non-zero values i;;v and i;;v of the -rotations in (3) are listed in Table1 . The rst index denotes position of the -rotation for a xed angle.... ..."

Cited by 4

### Table 11 - Ratio: equivalent variation / non zero bill

"... In PAGE 29: ... An interesting extension is to compare the willingness to pay and the bill actually paid for each traffic direction. Table11 gives the ratio between these two quantities. The last column indicates the percentage of customers who have a strictly positive demand for the ... ..."

### Table 1: Non-zero coe cients and scaling factors of the -rotations W , necessary for the prototype approximation in Figure 7 knl 0 1 2 3 4 5 6 7

1997

"... In PAGE 3: ....2.1. Example: N = 128, M = 8 Figure 7 shows the frequency response of a length-128 pro- totype for an 8-channel cosine-modulated lter bank de- signed with the QCLS-algorithm [3] and of the prototype allowing an e cient realization. The non-zero values i;v and i;v of the -rotations in (3) are listed in Table1 . The rst index denotes position of the -rotation for a xed angle.... ..."

Cited by 4

### Table 1: Number of non-zeros in matrix

"... In PAGE 4: ... The conjugate gradient methods turn out to have overhead per iteration in CG in relation to the rate of convergence. Moreover, the normal equation matrix surprisingly has less non-zero entries than the original matrix as show in Table1 below. This is due to the tensor product of the B-spline and the second order partial derivative that are within the profile of the B-spline subdivision matrix.... ..."

### Table 2: Estimated entropy of non-zero coe cients

2000

"... In PAGE 15: ... The entropy was estimated by computing the entropy of the non-zero coe cients (equation 14), which yielded lower estimated coding costs for the overcomplete bases compared to summing the individual entropies (equation 13). Table2 shows the estimated coding costs for the same learned bases and data sets used in table 1. The second column lists the cost of labeling the non-zero coe cients.... ..."

Cited by 145

### Table 4. Propagation of non-zero difference in internal state of the rekeying

2005

"... In PAGE 9: ... The larger the number of non-zero words used as input to the non-linear function, the more complex the resulting output. The key schedule of Dragon is designed so that after 12 rounds, even an initial difference of single word differ- ence is propagated to all words in the internal state (see Table4 ). Since there are 16 rounds, this is an ample margin to ensure an attacker is unable to determine the state contents after rekeying.... ..."

Cited by 7

### Table 3.5 - The distribution of non-zero strides.

1998

Cited by 25

### Table 3. Median number of non-zero coefficients

2005

"... In PAGE 13: ... The simulation results indicate that the elastic net dominates the lasso under collinearity. Table3 shows that the elastic net produces sparse solutions. The elastic net tends to select more variables than the lasso does, owing to the grouping effect.... ..."

Cited by 25

### Table 1. Comparison of Memory Requirement and Non-zero Density

2004

Cited by 6