### Table 1: Prime modulo set fragmentation.

2004

"... In PAGE 3: ...on-trivial 6.3%. Since we target the L2 cache, however, this fragmentation becomes negligible. Table1 shows that the percentage of the sets that are wasted in an L2 cache is small for com- monly used numbers of the sets in the L2 cache. The frag- mentation falls below 1% when there are 512 physical sets or more.... In PAGE 7: ...ure. The architecture is modeled cycle by cycle. Prime Numbers. The prime modulo function uses the prime number shown in Table1 . The prime displacement function uses a number 9 when it is used as a single hash- ing function.... ..."

Cited by 12

### Table 2 Parameters of four families of sets of three partitions, where q is a prime power congruent to 3 modulo 4

### Table 4: The prime factorization of the congruent pairs in Table 2 with the exponents taken modulo 2.

in 1

2002

"... In PAGE 14: ... Since we have to choose elements in such a way that their product is a square, we have to make sure that the exponent of each prime is even, in a and b together. Therefore the simpli ed Table4 is su cient for our algorithm. Let B denote the matrix of size n1 n2, where the elements are equal to the exponents modulo 2 of the prime factors, as given in Table 4.... In PAGE 15: ... It is also possible, especially for large numbers, that there are more solutions for the vector x. For example let x = [1; 1; 1; 0; 0; 0; 0; 1; 0; 0], which corresponds in Table4 to the 1st, 2nd, 3rd and 8th columns. It satis es also gcd(XY; N) = gcd(10240; 33) = 1, and gcd(X Y; N) = gcd(320 32; 33) = gcd (288; 33) = 3, resulting in factor 3.... ..."

### Table 1: Solutions for powers of primes (n 216) where if r is factored into powers of unique primes, r = s Y i=1 pai i

"... In PAGE 10: ... (Naturally, km?1 gt; 0 or else the cycle would be shorter than dm ? 1.) Table1 lists one set of weights K = fkm?1; : : :; k1; k0g for each power of a prime d where m 2 and dm 216. (No commas are needed because all weights are less than 10.... In PAGE 10: ... (No commas are needed because all weights are less than 10.) The solution from Table1 for n = 32 is illustrated in Figure 6 in condensed form, showing the rearrange- ment vectors fT1kg and fT2kg in radix{3. It can be seen that each vector is a permutation of Pn.... ..."

### TABLE 2: Value of a European put option at S a81 100 using Crank-Nicolson timestepping for linear, quadratic and cubic interpolation. The interpolation schemes are used to transfer data between the non- uniform S grid and the uniform log-spaced FFT grid. The input parameters are provided in Table 1. The convergence ratio R is de ned in equation (7.1). The exact solution is 3a87 149026. The number of points used for the FFT grid is 2a, where a is the smallest integer such that the number of nodes in the non-uniform S grid p a88 2a.

2005

Cited by 15

### Table 3: Poisson equation solution on a uniform mesh: in uence of the correction factor (with or without)

"... In PAGE 14: ...rids. The new feature is that directional coarsening has to be locally accounted for. Each basis functions gradients of the k + 1-level is multiplied by this matrix CI k;k+1 de ned by : CI k;k+1 = Rk i ?1 C k;k+1 0 0 C k;k+1 ! Rk i (10) where k + 1 denote the indice from the coarse level (coarse grid Gk+1 built from a ne grid Gk), Rk i is a rotation matrix of the coordinates (xi; yi) and C k;k+1, C k;k+1 are respectively a corrective factor of the basis function gradient ? !r I following and . For an isotropic mesh (see Figure 7 and 9, and Table3 ), these correction factors are equals C k;k+1 = C k;k+1 = Ck;k+1 and de ned together with the rotation matrix by : Ck;k+1 = p2 (Nk ? 1) (2Nk ? 1) ; Rk i = Id (11) where Nk is an approximation of the number of nodes in one direction (square root of the number of nodes nk of ne grid Gk (see [14])). For an anisotropic mesh (see Figure 11), these correction factors are put to unity in the stretching direction and Ck;k+1 in the orthogonal direction.... ..."

### Table 6: Government Phonology primes for some phones from the TIMIT set.

"... In PAGE 17: ... To represent the set of TIMIT segments, we allow three of the primes to be the head: A, I and U. Table6 shows the GP primes for some example phones from the TIMIT set. Table 7 shows the results for the GP system.... ..."

### Table 6: Government Phonology primes for some phones from the TIMIT set.

in Edinburgh

"... In PAGE 17: ... To represent the set of TIMIT segments, we allow three of the primes to be the head: A, I and U. Table6 shows the GP primes for some example phones from the TIMIT set. Table 7 shows the results for the GP system.... ..."

### TABLE 4: Value of a European digital put option using Rannacher timestepping and l2 projection. The input parameters are provided in Table 1. The convergence ratio R is de ned in equation (7.1). The exact solution is 0a87 854898 at S a81 90, 0a87 387153 at S a81 100, and 0a87 077923 at S a81 110. The number of points used for the FFT grid is 2a, where a is the smallest integer such that the number of nodes in the non-uniform S grid p a88 2a. Quadratic interpolation is used.

2005

Cited by 15