### Table 1: Parameters used in rst set of solutions of the TOKAMAK matrices.

1999

"... In PAGE 18: ...nd 83,842 nonzeros. The condition number is estimated at 1:9 109. The rst experiment compares single level ARMS preconditioners to a standard ILUT( ; p) preconditioner with two di erent levels of ll. The parameters in Table1 are the same for each simulation. The ll level for the grid transfer matrices is set to 50.... In PAGE 23: ...ed. Sys. Ind. Sets ILUT TOTAL ILUT(0,100) 19779 0 19779 0 3 632 937 3 632 937 ILUT(0,200) 19779 0 19779 0 6 934 486 6 934 486 ARMS-2 19779 12237 7542 2 488 896 0 7542 3716 3826 1 046 361 0 3826 2158 1668 559 183 0 1668 1123 545 251 411 0 545 403 142 83 114 17 146 4 446 111 Table1 0: Memory used in rst set of solutions of the RAEFSKY4 matrix. The convergence histories are shown in Figure 8.... In PAGE 24: ... The preconditioner parameters may be found in Table 11. General Intermediate Levels Last Level bsize 200 kmax;i 0 kmax;N 5 0 pi 100 pN 100 Table1 1: Parameters used in second set of solutions of the RAEFSKY4 matrix. The memory requirements for the preconditioners are shown in Table 12.... In PAGE 25: ... Sets ILUT Schur Comp. ARMS-1 ARMS-2 1 2 489 535 1 432 167 753 600 4 675 302 3 921 702 2 3 533 877 728 844 382 900 4 645 621 4 262 721 3 4 065 601 356 211 191 800 4 613 612 4 421 812 4 4 365 036 103 572 62 400 4 531 008 4 468 608 5 4 456 169 26 827 18 600 4 501 596 4 482 996 Table1 2: Memory used in second set of solutions of the RAEFSKY4 matrix. matrices, but that would also increase the factorization time, and the time required for each outer iteration.... In PAGE 26: ... This number drops to 81 for ILUT(0,200). General Intermediate Levels Last Level bsize 200 pi 100 pN 100 Table1 3: Parameters used in rst set of solutions of the BARTHT1A matrix. All three of the ILUT preconditioners tested require more memory than all of the ARMS preconditioners in this test.... In PAGE 26: ... The memory requirements listed in Table 14 include the sizes of the block LU matrices and the ILUT factorization of the last level. Method Total Memory Requirement ILUT(0,100) 2 763 711 ILUT(0,200) 5 352 233 ILUT(0,300) 7 805 162 1-level ARMS 2 188 913 2-level ARMS 2 300 116 3-level ARMS 2 367 746 4-level ARMS 2 396 588 5-level ARMS 2 408 088 Table1 4: Memory used in second set of solutions of the BARTHT1A matrix. Figure 10 compares the convergence history of the two methods.... In PAGE 27: ... This matrix shrinks as the number of levels increases. General Intermediate Levels Last Level bsize 200 kmax;i 0 kmax;N 5 0 pi 100 pN 100 Table1 5: Parameters used in second set of solutions of the BARTHT1A matrix. Memory Requirements Total Memory # Levels Ind.... In PAGE 27: ... Sets ILUT Schur Comp. ARMS-1 ARMS-2 1 1 410 769 778 144 427 060 2 615 973 2 189 913 2 1 999 912 300 204 172 800 2 472 916 2 300 116 3 2 238 557 129 189 75 300 2 443 046 2 367 746 4 2 346 530 50 058 32 600 2 429 188 2 396 588 5 2 391 218 16 870 14 700 2 422 788 2 408 088 Table1 6: Memory used in second set of solutions of the BARTHT1A matrix. The convergence histories are shown in Figure 11.... In PAGE 28: ... The full set of parameters used for the third set of tests on the BARTHT1A matrix is displayed in Table 17. General Intermediate level Last Level bsize 100 kmax;i 5 kmax;N 5 0 i 10?4 N 10?4 p0 300 pi 300 pN 100 Table1 7: Parameters used in third set of solutions of the BARTHT1A matrix. The preconditioners require slightly more memory, but convergence is attained in far fewer outer iterations, sometimes in less time.... In PAGE 29: ... Sets ILUT Schur Comp. ARMS-1 ARMS-2 1 1 767 389 784 288 798 045 3 349 722 2 551 677 2 2 987 222 391 094 1 398 678 4 776 994 3 378 316 3 3 676 705 182 967 1 688 780 5 548 452 3 859 672 4 4 053 399 71 919 1 812 680 5 937 998 4 125 318 Table1 8: Memory used in the third set of solutions of the BARTHT1A matrix. preconditioner consists of weighing the trade-o s involved between iteration on full and reduced systems.... In PAGE 30: ... General Last Level tolind 0.1 dimN 10 bsize 400 kmax;N 10 0 N 10?15 p0 400 pN 400 Table1 9: Parameters used in rst set of solutions of the BARTHT2A matrices. Memory usage is tabulated in Table 20.... ..."

Cited by 26

### Table VII: Tested Gas Mixtures, Gas Compositions, and Reactor

### Table 4: Description of the TOKAMAK matrices.

1999

"... In PAGE 12: ... Brown of Lawrence Livermore National Laboratory. Table4 shows some data on... ..."

Cited by 39

### Table 4: Description of the TOKAMAK matrices.

1999

"... In PAGE 12: ... Brown of Lawrence Livermore National Laboratory. Table4 shows some data on... ..."

Cited by 39

### Table 5.4 Description of the TOKAMAK matrices.

1999

Cited by 39

### Table 5.4 Description of the TOKAMAK matrices.

1999

Cited by 39

### Table 3. Character recognition tests on combined representation fusion matching.

"... In PAGE 16: ...able 2. Character recognition tests on TFV representation.............................................. 72 Table3 .... In PAGE 94: .... Return F. 5.6 Experimental Results on the Fusion Matching Table3 shows the character recognition test results using the combined representation fusion matching. Table 4 compares the recognition rates with those of the base representations.... ..."

### Table 3. Character recognition tests on combined representation fusion matching.

2001

"... In PAGE 16: ...able 2. Character recognition tests on TFV representation.............................................. 72 Table3 .... In PAGE 94: .... Return F. 5.6 Experimental Results on the Fusion Matching Table3 shows the character recognition test results using the combined representation fusion matching. Table 4 compares the recognition rates with those of the base representations.... ..."

### Table 3: Search test performance of the fusion system and its three components.

2002

"... In PAGE 9: ... Weights for the SDR-FUSION-SYSTEM were optimized us- ing the limited ground truth that was compiled for FTR+FV. As expected, this scheme led to Mean Average Precision (MAP) im- provements FTR+FV; more importantly, fusion gave performance improvements (35%) over our best single SDR system on the un- seen search test data (as shown in Table3 ). Note that simple post- evaluation changes in the video segment-to-shot mapping scheme improved the performance of the individual OKAPI systems (eg.... ..."

Cited by 8

### Table 1. Results for Score Fusion on the Entire Test Data

"... In PAGE 3: ... The pitch value and mismatch can still be accounted for, for example by calculating the fused score as a weighted average of the two scores, with weights that depend on the above indicator values. Table1 lists the DCF and EER obtained for different choices of constant and indicator- dependent weights 1 w and 2 w . The last entry in the table is for score fusion using constant weights obtained using the pDETAC optimization method described in [9].... ..."