### TABLE 2. Many degrees of freedom system with multiple equilibria; low-order statistics for the climate variable x and the first unresolved mode Y1.

2003

Cited by 4

### Table 3 shows the many-to-many relationship for one Key Practice. In order to ensure the absence of ambiguity and to simplify the task of translation of results, it may be necessary to explore the possibility of further decomposing the key practices, ideally to arrive at a one-to-many relationship to the elements of the reference model.

"... In PAGE 10: ... Thus, individual key practices may relate to multiple elements of the reference model; in addition, more than one key practice is frequently found to be related to a single 15504 element. Table3 - Relationship between CMM and 15504 basic elements... In PAGE 11: ... Table3 will demonstrate the possibilities. 6.... ..."

### TABLE 1. Many degrees of freedom system with stable periodic orbit; low-order statistics for the climate varibles x1 and x2;CTde- noted correlation time.

2003

Cited by 4

### Table 5.2 Comparison of weighted orderings. IF ll is how many thousands of nonzeros in the true inverse factors of the symmetric part, Pre-T is the time to compute the preconditioner, Its is the number of iterations for convergence, and Sol-T is the time taken by the iterations.

### Table 1 shows that the use of several Gauss-Seidel steps is far more e ective than just one step, as was shown in [6]. It also shows that the rate of convergence of ILU preconditioning is badly degraded in the red-black ordering, as expected, but that no degradation is experienced for Gauss-Seidel. Indeed, in many cases the number of iterations in the red-black ordering is considerably less than in the natural ordering.

"... In PAGE 1: ...5 15 1.7 55 53 58 58 Table1 : Gauss-Seidel as a preconditioner The results in Table 1 are for a nite di erence discretization of the convection- di usion equation ?(uxx + uyy) + (ux + uy) = f(x; y) (1.1) on the unit square with Dirichlet boundary conditions and = 5.... In PAGE 1: ...5 15 1.7 55 53 58 58 Table 1: Gauss-Seidel as a preconditioner The results in Table1 are for a nite di erence discretization of the convection- di usion equation ?(uxx + uyy) + (ux + uy) = f(x; y) (1.1) on the unit square with Dirichlet boundary conditions and = 5.... In PAGE 8: ...f at least one eigenvalue per restart cycle. Otherwise GMRES may stagnate. 3 Parallel Results The SOR iteration is implemented in the red-black ordering to achieve parallelism. As shown in Table1 and in more detail in [1], the use of the red-black ordering in the SOR-GMRES method does not degrade the rate of convergence as compared to the natural ordering. The grid is partitioned to the processors by strips, as shown in Figure 5.... In PAGE 15: ...5 0.90 Table1 0: Times (T), Speedup (S), E ciency (E) for (1.1) on the SP2, n1 = 301 The gradual degradation in performance occurs despite improvements in performance on each local grid comparable to those of the Paragon, where the cache utilization improved as the number of processors increased.... In PAGE 16: ...52 0.10 Table1 1: Times for varying grid shapes on a single SP2 node in Table 9) is signi cantly better than for two processors: Table 11 shows that the improvement in performance due to the change in the grid shape for two processors is negligible, and as a result the cost of communication causes the scaled speedup to be less than perfect. Next, we consider the e ects of a block partitioning.... In PAGE 16: ...057 0.78 Table1 2: Times for square local grids vs. square global grid on the SP2 For every case, partitioning by strips is signi cantly better (up to 18%) than using square local grids.... ..."

### Table 2 presents results in cases which violate at least one hypothesis of Lemma 4.1. In many cases, second order accuracy is retained, but in some cases it is lost. We omit the U?u vector error column, since it is always rst order, and the postprocessor from [11], since it was designed speci cally for three lines meshes. In their place, we provide both l2 and l1 convergence rates for the least squares postprocessor and for the normal uxes.

1998

"... In PAGE 11: ... These last two cases, as well as the cases in Table 1, are then repeated using the enhanced mixed method, which may be viewed as a generalization of cell centered nite di erences to triangles, using special (lower order) quadrature rules to simplify the solution of an extended mixed method formulation as described in [1, 2]. These cases are labeled with an (E) in Table2 . The normal uxes are less accurate, but the averaging e ect of the post-processor continues to maintain close to second order accuracy, at least in l2.... In PAGE 11: ... Computationally, however, the Least-Squares family of schemes are easily extended to meshes of tetrahedra, since no geometric regularity in the mesh is required by their construction. Numerically, we observe O(h1:5) convergence rates in l2, as indicated in Table2 . This indicates that while some form of superconvergence is present for tetrahedral meshes, and similarly for non-three-lines meshes of triangles, a more subtle analysis will be required to explain it.... ..."

Cited by 7

### TABLE 7 IMPORTANCE OF PERSONAL FACTORS TO HIRING DECISION

### Table 3 b) Without a priori knowledge of the shape of the solution In the most general case, we do not have a priori information on the shape of the solution. In order to nd the optimal solution, many paths must be tested which can have very various shapes.

### Table 2: Results on problems from Espresso benchmark suite For each example of a 3TBDD the number of states of the companion FSM is reported in the column denoted \# orig. states quot;. This number is always equal to the number of nodes of the 3TBDD plus one because a new node is added to the STG as explained in Section 7. The following two columns report the number of compatibles of the FSM (i.e. the cardinality of the set C(c)) and the number of compatibles after ltering as per Section 8.2 (i.e. the ones which are closed with respect to their l-class). This step reduces the number of compatibles of many order of magnitude. Then, after the number of primes, in column \# red. states quot; we reported the number of states of the 16

### Table 1. ManySim Simulation parameters

2007

"... In PAGE 7: ... T0 T1 L2 T2 T3 L1 L3 Memory Interconnect T4 T5 L2 T6 T7 L1 Figure 9. Evaluated CMP Environment Table1 summarizes the simulation configurations. As shown in the table, we model CMP architecture with a 3-level cache hierarchy and simple in-order cores.... ..."

Cited by 6