### Table 1 shows the convergence of the relative error in an adaptive iteration. It can be seen, that from about 10,000 grid points on, the ratio between error and estimate remains constant. The last column displays the grid e ciency, the product of the number of grid cells and the relative error. This number is about constant on all grids, displaying second order convergence on adaptive meshes. A typical computational grid is shown on the left Figure 3. Although the primal problem is symmetric, the weighted residual estimator induces a stronger grid re nement in the direction of interest. For comparison, a typical grid for a

1998

"... In PAGE 8: ...9e-2 7.70 231 Table1 : Comparison of relative errors and estimates From these numerical results, we conclude, that the norm k:kW is appropriate indeed to estimate stability for the radiative transfer equation. Furthermore, these results were obtained with a xed angular discretization of an odd number of intervals on S2, so no symmetry condition applies to the method.... ..."

Cited by 1

### Table 3 Results of leaps-and-bounds regression for best com- binations of the descriptorsa

"... In PAGE 3: ... Using variables V, S1, S2, S3, S4, S5, S6, S7 and K Because the above equation is not sufficiently strong to allow confident prediction, thus, the indicated variable K was added. The leaps-and-bounds regression results are shown in Table3 . This Table revealed that to put together the correlation coefficients, significance tests, and standard deviations, 7-variable combination including variables 1, 2, 3, 5, 6, 7 and 9 is the best one.... ..."

### Table 2 Results of leaps-and-bounds regression for best com- binations of the descriptorsa

"... In PAGE 3: ... In the situation of this study, we have 35 samples, so M can not be great than 7. Table2 shows the results of leaps-and-bounds regres- sion for the best combinations of the 8 descriptors. For example, the best one-variable selection is 1 (V).... ..."

### Table 1.2: Equational rules for -algebras

in Functionality, Polymorphism, and Concurrency: A Mathematical Investigation of Programming Paradigms

1997

Cited by 3

### Table 6: IKSP(v; 3)s

"... In PAGE 7: ... This completes all cases. 2 4 KSPs with v 1 (mod 6) Table6 gives IKSP(v; 3)s for a number of small orders, but the technique is somewhat limited in this case when v 7 (mod 12) since then t is even and the resulting short di erence t 2 prevents the construction apos;s application. Table 7 remedies this by employing two base parallel classes in Zt=2 f0; 1; 2; 3g in the case v = 31; here each base parallel class provides t 2 of the parallel classes when developed modulo (t=2; ?).... ..."