### Table 2. Results of Branch-and-Bound for Various Partitioning Schemes Two Rectangles Four Rectangles Triangles and Rectangles

"... In PAGE 25: ... This triangular partitioning scheme was compared against a two-rectangle partitioning scheme, where the rectangle was divided into two subrectangles by bisecting the longest edge and a four-rectangle scheme obtained by bisecting the original rectangle along both edges. The results of the computational experiment are given in Table2 . Figure 9 shows a performance profile (see [10] for an explanation of performance profiles) of the number of nodes of the branch-and-bound tree, and Figure 10 shows a profile of the CPU time used by each of the three partitioning methods.... ..."

### Table 4: Averaged shelling for the T ubingen triangle tiling.

"... In PAGE 5: ... Due to the dihedral D10 symmetry of the window, the result is the same for the entire D10 orbit of x?, thus it is su cient to consider one representative, and multiply the covariogram by the length of the orbit. The result for the T ubingen triangle tiling is given in Table4 , which contains all possible radii r 3, completing and extending a previously published table[7] where one possible radius was missed. Here, t2 = (r2) = (xx) is the squared length of x?, and is algebraic conjugation in the eld Q( ), which maps p5 7! p5, hence ( ) = 1= = 1 .... ..."

### TABLE I PERCENTAGES OF TRIANGLES THAT FIT IN A GIVEN TILE SIZE (PIXELS).

2002

Cited by 1

### TABLE I PERCENTAGES OF TRIANGLES THAT FIT IN A GIVEN TILE SIZE (PIXELS).

2002

Cited by 1

### Table 1: Comparison of #0Cve algorithms on random CSPs with D=3. Eachnumber is the mean of 2,000 solvable

"... In PAGE 4: ...that correspond to the phase transition #28Mitchell, Sel- man, amp; Levesque 1992#29 as determined by BJ+DVO. In Table1 and Table 2 we display the results of ex- perimenting with particularly big and di#0Ecult problem instances, using a collection of algorithms that were ob- served to be superior over various other variants in our earlier experiments. The tables displayaverage cpu seconds.... ..."

### Table 2: Comparison of #0Cve algorithms on random problems with D=6. Eachnumber is the mean of 2,000 solvable

"... In PAGE 4: ...that correspond to the phase transition #28Mitchell, Sel- man, amp; Levesque 1992#29 as determined by BJ+DVO. In Table 1 and Table2 we display the results of ex- perimenting with particularly big and di#0Ecult problem instances, using a collection of algorithms that were ob- served to be superior over various other variants in our earlier experiments. The tables displayaverage cpu seconds.... ..."

### Table 1. Number of triangles transferred as a function of the tile size for each benchmark.

"... In PAGE 6: ... Therefore, a better indication of an appropriate tile size is the number of triangles sent to the rasterizer. Table1 depicts the number of triangles transferred to the rasterizer for various tile sizes. As explained in Section 3, this data usually dominates the data front component of the total external data.... ..."

### Table 2. Number of triangles with a given number of tangents, out of 5 000 000 randomly constructed triangles

2005

Cited by 3

### Table 1: Random variables in a run-ofi triangle.

in Abstract

2005

"... In PAGE 2: ... We refer to England amp; Verrall (2002) for an overview. The methods discussed by these authors are framed within the context of a run-ofi triangle like the one in Table1 . The random variable Yij (for i; j = 1; : : : ; t) denotes the claim flgure for year of origin (arrival or incurral year) i and development year j, made up by aggregating the individual claims corresponding with this (i; j) combination.... In PAGE 2: ...Table 1: Random variables in a run-ofi triangle. The present literature on loss reserving only contains techniques based on summary triangles like the one in Table1 . However, some authors recently suggested to leave the track of aggregate claim flgures.... In PAGE 3: ... The number of claims in arrival year i is denoted by ni and t(i; k) denotes the development year of the last observation for the kth claim from arrival year i. As in Table1 , the flgures represented by the random variables can be for instance incremental, cumulative or incurred payments or loss ratios. For every claim in a unit record data set, repeated measurements (e.... In PAGE 7: ... In further work, the mixed models presented here can be extended to other distributional frameworks and can be adopted to model zero or negative incremental payments or censored data by using two-parts models based on generalized linear mixed models (GLMM). Applied to a run-ofi triangle like the one in Table1 , the general lognormal regression 1SAS is a commercial software package (for details see http://www.... ..."

### Table 3: The expected area of the bounding rectangle for a random sample and the expected length of a side of the bounding rectangle for the single window algorithms and the extended window algorithms. This shows the bias of the extended window algorithms toward smaller bounding rectangles.

1987

"... In PAGE 20: ... Thus, a substantial bias toward smaller bounding rectangles could lead to poorer ts. Equations determining the amount of bias are derived in Appendix B (see equations 14 through 17) and representative values are shown in Table3 . As seen in this table, although the bias toward smaller windows exists, it is not as severe as might be feared.... ..."

Cited by 1