Supplementary Materials for Optimizing Two-level Supersaturated Designs by Particle Swarm Techniques
|
|
- Eleanor Page
- 5 years ago
- Views:
Transcription
1 Supplementary Materials for Optimizing Two-level Supersaturated Designs by Particle Swarm Techniques Frederick Kin Hing Phoa Institute of Statistical Science, Academia Sinica and Ray-Bing Chen Department of Statistics, National Cheng Kung University and Weichung Wang Institute of Applied Mathematical Sciences, National Taiwan University and Weng Kee Wong Department of Biostatistics, University of California at Los Angeles September 5, Numerical Studies for the choices of q LB and q GB in the MIX Operation We recommend q LB = m/3 or m/4 and q GB = m/6 in the MIX operation. This recommendation is based on a few simulation studies that used SIBSSD to find E(s 2 )-optimal SSDs. For space consideration, we describe one here where we consider finding an optimal SSD. For each combination of q LB and q GB, we ran each search for the LB and GB SSDs 100 times. The number of iterations and the approximate time for finding the Supported by the Career Development Award of Academia Sinica (Taiwan) grant number 103-CDA- M04 and National Science Council (Taiwan) grant number M MY3. Supported by the National Science Council (Taiwan) grant number M MY2 and the Mathematics Division of the National Center for Theoretical Sciences (South) in Taiwan. Supported by the National Science Council (Taiwan) and the Taida Institute of Mathematical Sciences. Supported by National Science Council, Mathematics Research Promotion Center grant number and the Ministry of Education (Taiwan) Top University Project to the National Cheng Kung University. 1
2 E(s 2 )-optimal SSD were recorded. As mentioned before, we assume that q LB > q GB to avoid premature convergence of all SSDs to the GB SSD. Table S.1 reports the simulation results. The four numbers in each box are the mean number of iterations, the standard deviation of the number of iterations required, the average time required in the 100 trials and its corresponding standard deviation. All simulations were done in R version 2.15 on a laptop equipped in Intel Core i7-2820qm CPU at 2.30GHz. The average time is obtained from the R function system.time. There are several observations from Table S.1. First, if the q s are set to be too small, more iterations are required to obtain the E(s 2 )-optimal SSD because information from the LB and GB SSDs are not available to the SSDs in the next iteration. In the extreme case, when we have q GB = q LB = 0, the whole procedure becomes a random search. Second, if the q s are set to be too large, extra time is required for the computational steps for the column exchanges. In addition, large values of q s may mean that the SSDs are more likely to be trapped in a local optimum and the lack of random columns replacement makes the escape from the local optimum more difficult. Simulation results shown in Table S.1 suggest the choices for q LB = 3 or 4 and q GB = 2 respectively, which match our recommendation. SIBSSD did well with q LB = [m/3] and q GB = [m/6] for the local and global searches of SSDs reported in the paper, where [x] is the positive integer as large as x. The average number of iterations required by SIBSSD shown in Table S.1 are not large, but the CP algorithm is faster than SIBSSD in terms of the time required for convergence. However, short convergence time of the CP algorithm does not mean that the final SSD is guaranteed to be optimal or near optimal either. This can be demonstrated in a simple simulation study replicated 100 times to find the E(s 2 )-optimal SSD. 2
3 Table S.1: Simulation results for different choices of q LB and q GB when m = 12 and N = 10. The 4 numbers in each cell are: the mean number of iterations, the standard deviation of the number of iterations, the average time (in seconds) and the standard deviation of the time (in seconds) required in the 100 trials. Choice of q GB for GB SSD Choice of q LB for LB SSD
4 2 SIBSSD-Generated Design Tables Tables S.3, S.4 and S.5 show the SIBSSD-generated SSDs, which are comparable to those reported in Bulutoglu and Cheng (2004) for N = 10, 12, 14, 16. These tables include the dimension of SSDs, their E r -efficiencies calculated from Bulutoglu and Cheng (2004) and the column indexes of these SSDs. The last column in the three tabless is the column indexes of our best SSDs obtained via SIBSSD. Each number in a column index is the decimal representation of a binary (two-level) column. Table S.2 demonstrates the correspondence between the column index and the column level settings of the SSD. To transform a decimal number (column index) 203 into its binary representation (two-level factor settings), we first find the binary number of 203, which is Then we map every (0, 1) digit into ( 1, +1) entry and obtain a binary column ( 1, 1, +1, +1, 1, 1, +1, 1, +1, +1) shown in the first column of Table S.2. Table S.2: SSD listed in Table 3. Column Index SSD Columns Practical Uses in Screening Experiments The SIBSSD-generated SSDs can provide good alternatives to some commonly used SSDs in scientific research and in practice. Consider, for example, the well known SSD D Lin proposed by Lin (1993) in Table S.6 and the SIBSSD-generated E(s 2 )-optimal SSD D P shown in Table S.4. The value of E(s 2 ) for D P is , which is slightly smaller than the value of from D Lin. It is instructive to compare designs based on additional characteristics and in this case, we further compare these two designs based on their ability to identify active factors. For this purpose, we set up a simulation study to ascertain capabilities of these two designs to identify active factor in a screening experiment. Among all variable selection methods for SSDs available in the literature, we choose to use the Stepwise Response Refinement Screener (SRRS) proposed by Phoa (2014) to identify active factors because it is one of the efficient and consistent methods for selecting active factors using SSDs. The simulation results in Phoa (2014) showed that the SRRS method performs well in SSD variable selection when compared to several commonly used methods in the literature. We set up our simulation similar to Example 4 of Phoa (2014) using ideas from Marley and Woods (2010) who suggested that the number of runs should be at least three times 4
5 Table S.3: SIBSSD-generated SSDs and their Er-efficiencies calculated from Bulutoglu and Cheng (2004) for N = 10 and 12. N m Er Column Indexes of SSDs found by the SIBSSD algorithm % (47, 179, 236, 286, 369, 436, 454, 661, 805, 905, 930) % (94, 199, 403, 412, 535, 601, 653, 692, 714, 782, 837, 976) % (920, 286, 838, 422, 810, 307, 466, 820, 535, 124, 397, 865, 341) % (203, 218, 271, 453, 468, 535, 604, 665, 710, 737, 842, 849, 908, 914) % (117, 218, 279, 398, 454, 527, 598, 690, 709, 812, 835, 856, 866, 913, 916) % (87, 124, 174, 434, 460, 481, 551, 618, 661, 665, 740, 752, 809, 820, 850, 902) % (117, 158, 167, 205, 285, 316, 331, 342, 422, 466, 535, 570, 590, 724, 775, 868, 913) % (91, 107, 182, 248, 286, 348, 358, 426, 496, 570, 572, 614, 716, 722, 817, 850, 872, 906) % (721, 355, 794, 376, 395, 569, 341, 94, 838, 434, 472, 812, 992, 234, 397, 841, 716, 109, 436) % (614, 316, 174, 662, 434, 355, 218, 484, 460, 810, 744, 395, 122, 782, 55, 79, 342, 167, 930, 451) % (866, 542, 835, 527, 682, 214, 626, 103, 283, 902, 364, 844, 203, 689, 601, 314, 968, 557, 234, 425, 220) % (803, 722, 117, 849, 179, 696, 203, 992, 376, 589, 569, 614, 744, 107, 302, 677, 118, 542, 611, 244, 818, 602) % (618, 484, 908, 738, 364, 597, 451, 199, 331, 472, 117, 310, 186, 782, 865, 710, 852, 850, 728, 929, 205, 557, 913) % (604, 810, 47, 661, 905, 283, 844, 301, 357, 91, 121, 460, 817, 182, 572, 790, 440, 682, 611, 310, 527, 632, 157, 677) % (908, 696, 299, 182, 61, 602, 880, 286, 614, 535, 818, 932, 837, 103, 682, 403, 899, 466, 362, 206, 782, 555, 601, 316, 662) % (796, 422, 211, 914, 206, 481, 692, 618, 121, 397, 55, 849, 372, 709, 364, 740, 803, 838, 434, 936, 342, 186, 472, 752, 714, 482) % (677, 279, 539, 117, 457, 961, 299, 468, 880, 541, 812, 331, 789, 109, 179, 412, 929, 824, 614, 185, 728, 589, 647, 398, 236, 602, 369) % (110, 419, 211, 721, 841, 866, 852, 229, 570, 557, 809, 654, 122, 440, 696, 205, 458, 436, 992, 902, 714, 236, 348, 628, 611, 283, 662, 728) % (426, 211, 410, 976, 327, 737, 803, 213, 662, 121, 236, 626, 454, 618, 539, 376, 174, 242, 94, 203, 614, 58, 955, 372, 542, 692, 714, 185, 167) % (710, 213, 59, 653, 422, 787, 342, 234, 930, 976, 472, 566, 651, 428, 218, 334, 395, 620, 227, 181, 542, 333, 920, 188, 214, 632, 654, 31, 314, 436) % (854, 873, 1251, 1944, 2265, 2362, 2508, 2605, 2800, 2947, 3409, 3658, 4000) % (423, 633, 670, 1111, 1227, 1340, 1992, 2284, 2382, 2457, 2898, 3130, 3284, 3597) % (441, 629, 795, 854, 1115, 1492, 1677, 1714, 1896, 2110, 2499, 2776, 3185, 3356, 3857) % (311, 858, 963, 1295, 1464, 1477, 1579, 1644, 2203, 2409, 2854, 2956, 3158, 3402, 3490, 3857) % (380, 485, 603, 867, 918, 1079, 1142, 1258, 1295, 1490, 1848, 2234, 2355, 2382, 3409, 3668, 3874) % (119, 500, 729, 754, 873, 1258, 1457, 1596, 1827, 2107, 2412, 2488, 2499, 2725, 2866, 3185, 3808, 3920) % (571, 742, 977, 1197, 2165, 2227, 2254, 2410, 2473, 2620, 2695, 2761, 2844, 3226, 3312, 3335, 3657, 3682, 4000) % (411, 627, 718, 853, 934, 1103, 1268, 1386, 1677, 1746, 1814, 2172, 2262, 2499, 2599, 2892, 3143, 3350, 3610, 4032) % (245, 619, 726, 828, 934, 965, 1118, 1319, 1506, 1512, 1678, 1805, 1873, 1874, 2375, 2382, 2866, 3016, 3476, 3684, 3715) % (231, 287, 629, 903, 996, 1244, 1325, 1458, 1475, 1713, 1813, 1874, 2358, 2417, 2513, 2710, 2953, 3143, 3221, 3492, 3619, 3908) % (543, 2857, 2103, 2769, 882, 3672, 715, 2978, 869, 700, 1638, 1713, 1940, 2758, 237, 3628, 2396, 3653, 2292, 2837, 2702, 634, 3123) % (3682, 685, 2529, 3207, 3404, 1272, 715, 2474, 359, 1307, 1954, 1992, 814, 1841, 3972, 2889, 2620, 3369, 2458, 3595, 1134, 854, 1477, 2914) % (603, 741, 1382, 3521, 2892, 2095, 3192, 1678, 3668, 2851, 3143, 2837, 3010, 3626, 1355, 1649, 1805, 3745, 237, 2457, 2635, 1833, 2412, 3213, 2792) % (374, 1198, 3622, 3162, 1203, 1295, 2858, 1938, 1859, 2247, 730, 3552, 2611, 1641, 490, 1638, 2347, 3356, 1145, 700, 3211, 95, 3186, 2762, 1818, 435) % (3118, 2375, 3858, 1241, 543, 1738, 630, 1127, 3276, 1925, 2199, 1310, 1460, 3668, 474, 3250, 686, 1478, 2172, 3040, 1649, 3732, 2259, 437, 860, 1394, 3353) % (3850, 2953, 1830, 2898, 3476, 2757, 2343, 3633, 1179, 2620, 1678, 444, 2738, 1000, 3496, 1641, 1925, 3180, 3612, 1841, 1748, 698, 2725, 797, 1520, 3225, 2904, 3808) % (207, 3164, 413, 2835, 3126, 3717, 1637, 1457, 3609, 599, 1746, 1355, 3722, 1055, 2829, 1806, 952, 732, 2665, 2394, 1873, 3225, 2755, 3524, 1195, 2199, 3521, 1926, 1752) % (3012, 238, 2364, 3238, 3342, 1381, 3636, 606, 2155, 685, 807, 3808, 2261, 2883, 498, 3688, 2954, 2866, 2412, 3158, 1520, 1932, 3459, 1890, 2758, 1358, 2520, 3242, 438, 1690) % (3347, 2860, 3976, 3605, 1938, 3619, 1086, 1477, 3312, 441, 423, 1716, 854, 3428, 1309, 2206, 3214, 2964, 2418, 1834, 1446, 1372, 2481, 1731, 3466, 1209, 1000, 2266, 795, 694, 207) % (3721, 3366, 854, 694, 2508, 2885, 1880, 2172, 3666, 1637, 2247, 2992, 1445, 915, 1309, 2617, 867, 3846, 1131, 629, 1678, 287, 1836, 811, 748, 1685, 1266, 1520, 2103, 3636, 686, 1118) % (1611, 3115, 1268, 2738, 1457, 3640, 573, 2898, 3591, 2417, 1817, 1427, 3605, 2203, 938, 1746, 1421, 3171, 3350, 1086, 2446, 4000, 679, 1251, 1896, 3748, 807, 2761, 725, 1690, 819, 3969, 2233) % (2460, 2788, 190, 3621, 1229, 3095, 423, 3718, 2390, 2095, 821, 2381, 599, 2699, 2530, 1131, 1806, 903, 1268, 574, 3366, 2737, 2666, 1836, 3377, 2652, 882, 365, 3908, 741, 2984, 1986, 2278, 1212) % (492, 1738, 1904, 854, 2788, 3522, 1272, 3681, 2738, 2142, 725, 371, 1127, 3880, 3221, 1925, 1148, 1458, 2473, 2868, 1817, 3397, 1436, 497, 1764, 573, 3672, 2265, 3024, 3252, 1683, 3126, 3412, 798, 1422) % (2649, 1889, 980, 952, 238, 2343, 459, 2892, 2977, 2457, 1818, 573, 2890, 1706, 783, 371, 2707, 2674, 2454, 2755, 249, 2732, 543, 2410, 966, 2866, 3226, 2000, 2396, 3465, 3653, 2292, 2984, 3380, 3688, 3718) 5
6 Table S.4: SIBSSD-generated SSDs for N = 14 and their Er-efficiencies as measured by the lower bounds from Bulutoglu and Cheng (2004). N m Er Column Indexes of SSDs found from the SIBSSD algorithm % (2425, 4813, 7076, 7632, 8677, 9628, 9968, 10838, 11913, 12467, 13160, 14396, 14538, 15121, 15429) % (9107, 10271, 13016, 14898, 15633, 4879, 6371, 5302, 3733, 5211, 2522, 4537, 7818, 6996, 2681, 14533) % (1453, 2419, 5415, 5745, 6819, 6956, 8933, 9138, 9771, 10567, 11444, 12128, 12776, 13077, 13414, 14393, 15745) % (823, 1772, 3952, 4565, 5167, 5436, 6019, 6761, 7073, 7253, 8569, 10053, 11429, 12902, 12980, 13792, 13849, 15116) % (1459, 1739, 2279, 3358, 3670, 4219, 4566, 5484, 7309, 7842, 8892, 9263, 9429, 10208, 10698, 13063, 13414, 13466, 15441) % (830, 1149, 1521, 1750, 2255, 3498, 4715, 5017, 5415, 5517, 6082, 6323, 7701, 8539, 8599, 8871, 9897, 12035, 12772, 13466) % (1486, 2477, 3865, 4159, 5042, 6734, 7273, 8917, 9068, 9338, 10467, 10550, 11019, 11420, 12633, 12941, 13575, 13736, 14490, 14676, 15908) % (1822, 2207, 2801, 2918, 3881, 4663, 5003, 5229, 5242, 6826, 8427, 8828, 9035, 9779, 9900, 10299, 11430, 12583, 13232, 14648, 14861, 15938) % (1207, 1479, 1964, 2426, 2774, 3181, 4726, 5041, 5275, 5368, 5678, 6439, 6556, 7876, 8682, 8847, 9588, 10428, 11825, 12381, 12517, 13590, 15522) % (1815, 3018, 4596, 4717, 5347, 6471, 7331, 8622, 8819, 9125, 9543, 9929, 10299, 10453, 11492, 12649, 13010, 13187, 13454, 13656, 14737, 14861, 14918, 16160) % (13861, 8072, 15170, 14561, 15506, 11545, 5489, 699, 11880, 5722, 3875, 6964, 12590, 6685, 9107, 6443, 10302, 4600, 1359, 10152, 12875, 7307, 12685, 5798, 13369) % (5421, 7702, 14382, 9912, 5964, 3685, 13081, 9373, 6825, 5827, 13988, 3560, 3230, 10829, 9126, 3896, 15745, 9333, 11555, 13418, 15696, 12164, 10058, 14960, 1843) % (942, 1207, 1630, 2020, 2485, 2669, 3613, 3875, 4455, 4473, 5048, 5182, 5771, 5901, 6762, 6934, 8889, 8990, 9043, 9515, 9580, 10414, 12839, 12997, 13876, 14605, 16160) % (251, 1516, 1749, 2936, 3251, 3662, 3785, 3847, 5003, 5074, 5181, 5678, 6247, 6826, 8798, 8871, 9571, 9786, 10121, 10383, 10857, 11172, 11546, 12621, 14048, 14492, 15554, 15889) % (501, 1359, 1402, 1756, 2703, 3041, 4318, 5073, 5347, 6082, 6550, 6986, 7592, 9099, 9126, 9625, 9960, 10482, 10586, 11459, 12100, 12587, 12748, 12908, 13137, 13383, 13970, 14537, 15748) % (956, 1199, 2921, 3290, 4222, 4579, 5930, 6429, 6791, 6962, 7648, 8654, 8821, 9123, 9656, 10287, 10542, 10649, 11204, 11315, 12521, 12660, 12691, 12954, 13581, 13897, 14086, 14666, 15016, 15524) % (10026, 8678, 11057, 13589, 10697, 15524, 3988, 11845, 8855, 1837, 13912, 5795, 12972, 3215, 9625, 6938, 7073, 4461, 8892, 1990, 13961, 14730, 11564, 5616, 15395, 3816, 14086, 12224, 14924, 5061, 9059) % (8636, 1679, 5548, 5872, 3992, 13081, 9433, 6805, 6539, 13962, 5179, 12065, 5089, 14506, 14441, 14870, 13637, 7632, 463, 4954, 7197, 6956, 11596, 15746, 3379, 3659, 14089, 7756, 10984, 11141, 5797, 6361) % (2143, 15875, 1849, 10865, 6857, 1869, 3878, 7764, 9884, 11370, 12839, 1987, 2742, 9457, 9188, 10637, 7333, 12629, 12908, 11202, 9367, 5478, 635, 1765, 4766, 9295, 4523, 14104, 5362, 7880, 11436, 2428, 8938) % (10905, 9188, 6926, 15528, 3796, 14104, 10446, 12950, 14915, 14396, 954, 11814, 9613, 15233, 5864, 12629, 11184, 3223, 8068, 13762, 13861, 5779, 11504, 8909, 13528, 4285, 6772, 6625, 3928, 6857, 14738, 8819, 11541, 11787) % (3306, 10806, 5830, 5179, 6550, 14444, 15010, 8861, 4794, 1852, 7333, 9359, 7694, 7576, 4947, 15619, 1327, 4999, 3017, 11787, 14120, 11924, 10178, 10650, 4001, 12716, 9650, 7768, 14216, 13521, 10574, 13652, 1246, 3861, 2020) % (10673, 2846, 12035, 11480, 7976, 6741, 6475, 1437, 12636, 14505, 6064, 12698, 9271, 13666, 4339, 9158, 11922, 702, 2363, 6866, 3430, 3537, 10872, 13402, 15500, 7284, 9080, 1770, 14128, 10724, 7446, 6081, 15236, 14883, 9045, 13837) % (9326, 3482, 9020, 6733, 9877, 10724, 16064, 1001, 3638, 12920, 14086, 9557, 3629, 11043, 12168, 8910, 5030, 3028, 6000, 11075, 1898, 13729, 9811, 1703, 15460, 5451, 3909, 10467, 2422, 9676, 11321, 14862, 12567, 16160, 9005, 6429, 3470) % (9422, 5785, 1882, 12815, 14930, 3909, 10523, 14785, 4503, 7562, 4476, 13609, 10680, 6251, 9123, 13521, 14988, 223, 3761, 8921, 4835, 13106, 7060, 9188, 4969, 13650, 13876, 8128, 2746, 11377, 6268, 1821, 11084, 15881, 3367, 13484, 14613, 14855) % (14876, 9515, 10693, 7073, 15497, 11938, 2796, 6000, 3877, 4207, 13094, 7219, 1214, 8825, 7693, 10556, 13540, 4537, 13194, 9877, 2231, 1933, 8751, 11576, 10835, 3862, 15397, 8599, 2414, 15010, 13638, 3659, 7310, 6423, 14665, 3523, 997, 11492, 11146) % (3642, 2477, 5287, 10767, 14993, 12068, 2802, 9785, 9626, 6460, 8429, 847, 1756, 3159, 7310, 11403, 11349, 8814, 8606, 5895, 3980, 8629, 14489, 11604, 1270, 985, 2359, 13078, 15044, 4033, 6520, 5213, 2591, 11594, 10595, 2515, 5481, 14426, 10003, 12742) % (1694, 3537, 13736, 8351, 2510, 5972, 15044, 13187, 9698, 10550, 4533, 998, 13398, 10992, 9052, 7718, 13389, 14599, 13492, 10115, 10901, 14224, 11858, 3753, 4915, 1821, 9068, 4826, 4837, 8315, 1214, 6938, 2517, 9005, 14744, 11660, 4600, 12491, 6798, 1479, 7497) % (9123, 4507, 889, 9786, 7400, 8909, 4207, 10481, 2223, 15619, 1483, 3373, 2518, 7009, 12724, 1143, 12445, 10541, 10457, 5036, 1510, 15010, 3733, 9031, 13351, 3467, 4339, 10302, 6486, 2789, 11462, 15464, 3635, 5301, 6671, 2746, 7701, 9649, 9870, 12379, 14979, 10550) 6
7 Table S.5: SIBSSD-generated SSDs for N = 16 and their Er-efficiencies as measured by the lower bounds from Bulutoglu and Cheng (2004). N m Er Column Indexes of SSDs found from the SIBSSD algorithm % (891, 5295, 10429, 11667, 14280, 25693, 28914, 29033, 34292, 39129, 41445, 41671, 44138, 45342, 46641, 50571, 58040) % (1455, 3167, 7020, 12955, 14398, 18041, 21686, 22937, 27427, 30026, 36634, 37235, 41421, 48169, 51434, 53775, 56581, 57660) % (12647, 11506, 30372, 5694, 14923, 3953, 43798, 52263, 43685, 48424, 61491, 46481, 6333, 40134, 9615, 42108, 27677, 34475, 38485) % (1515, 6494, 13005, 16164, 18374, 20791, 25724, 27089, 34645, 35997, 38130, 41383, 41848, 45270, 51684, 51787, 53357, 54168, 58638, 62785) % (5246, 7822, 8891, 10744, 15784, 19683, 20903, 21464, 26782, 30249, 34740, 36123, 37323, 39093, 42189, 47714, 50424, 51882, 55353, 57646, 62610) % (17333, 21865, 18682, 39904, 59683, 29578, 11689, 10086, 12755, 52614, 59073, 34618, 32290, 58608, 12120, 3811, 24464, 52041, 38539, 47665, 38227, 46504) % (44357, 26403, 64096, 50886, 33769, 39207, 19285, 59794, 11180, 13134, 41623, 61485, 16005, 40714, 58652, 37812, 3135, 23193, 44600, 43211, 13808, 54665, 45843) % (2806, 7470, 11155, 12661, 14130, 18221, 21338, 25541, 26142, 27747, 27985, 31892, 36633, 36678, 41387, 42549, 43356, 46659, 50672, 51255, 53527, 57466, 58758, 64288) % (7725, 10938, 12761, 13030, 15893, 17207, 19700, 19770, 22742, 24386, 25209, 25743, 27459, 31004, 34927, 35733, 38490, 44115, 44960, 45366, 53937, 55946, 58902, 59980, 63523) % (24226, 26220, 45682, 40152, 52592, 27846, 22165, 54666, 48897, 42697, 29779, 34486, 51371, 1523, 53985, 31632, 43189, 4589, 10168, 57812, 54332, 6750, 49947, 46374, 28858, 27705) % (5606, 56389, 32385, 22830, 47938, 15629, 3631, 15272, 36554, 46644, 54826, 40280, 63622, 41197, 36772, 53673, 37767, 47212, 59148, 28002, 19164, 13470, 30432, 44083, 58762, 51636, 40120) % (759, 13669, 15507, 6861, 3485, 41055, 59557, 22649, 39380, 10744, 36405, 31260, 37054, 50129, 54476, 46617, 37771, 51479, 44107, 45357, 3886, 25485, 44740, 52393, 39777, 21831, 31363, 51790) % (2782, 42883, 27749, 52009, 23948, 10730, 12623, 49637, 44868, 27179, 36043, 62152, 59670, 31618, 26009, 18162, 59601, 7137, 10909, 32321, 11760, 22867, 17351, 12981, 25464, 1883, 25767, 26444, 20245) % (1901, 23244, 51847, 8081, 29781, 43293, 26266, 49401, 60864, 38285, 6379, 23732, 16855, 29577, 19755, 52056, 43685, 54180, 52785, 2550, 30983, 40770, 54318, 21299, 1950, 20045, 37463, 14908, 26468, 55580) % (46914, 34524, 21969, 12053, 15088, 8439, 3019, 39318, 5031, 51315, 15418, 1530, 24133, 17270, 39257, 59796, 39526, 3870, 36529, 11113, 13684, 23352, 7341, 45992, 29211, 37181, 21212, 23970, 50993, 26988, 14801) % (57525, 50858, 48004, 19180, 31257, 25330, 11578, 36438, 12910, 19239, 52284, 31842, 27783, 58697, 30500, 55776, 11224, 54042, 59266, 33135, 19251, 43689, 43378, 18891, 25020, 2033, 7354, 20894, 45227, 11662, 59470, 12503) % (36069, 15692, 51531, 38489, 5561, 46469, 57532, 29409, 39284, 53462, 40106, 50471, 4719, 47155, 9587, 19669, 41433, 42526, 63513, 13455, 41581, 27880, 34546, 54060, 39563, 61890, 54377, 48960, 43937, 4602, 62562, 22321, 14391) % (36409, 51681, 27733, 48736, 3452, 27818, 61740, 58800, 12785, 56082, 54421, 20420, 6390, 5927, 54498, 26919, 48452, 50515, 13722, 32385, 57910, 22709, 21611, 8112, 47379, 29304, 23582, 8919, 47245, 21337, 51418, 11379, 31554, 2987) % (6462, 13268, 61994, 58646, 43633, 9115, 26797, 25691, 38130, 10085, 11506, 18408, 34510, 34215, 33469, 53955, 58800, 51653, 40217, 43922, 3735, 20259, 49499, 61064, 44364, 33273, 46881, 51828, 43850, 23010, 17214, 8088, 20305, 10667, 36458) % (13901, 55905, 23864, 4671, 26741, 20058, 57689, 20935, 25398, 27726, 30596, 18845, 18289, 56341, 13651, 31378, 39254, 60163, 15117, 29034, 14521, 36139, 25779, 15824, 53466, 64515, 3811, 46196, 41689, 10726, 58540, 34199, 43574, 44824, 45355, 54240) % (29036, 43490, 43852, 22923, 17273, 52833, 12208, 25306, 62744, 38852, 2973, 3534, 55822, 37095, 32386, 14934, 7467, 43070, 29868, 37810, 47257, 1726, 15141, 52532, 26375, 44681, 23272, 55700, 57517, 13405, 45866, 58245, 38729, 14801, 42442, 5052, 15588) % (43802, 39115, 33526, 28974, 19578, 14580, 40241, 25049, 5789, 53596, 41083, 14940, 10216, 53865, 62546, 13155, 60041, 46504, 21051, 35692, 57653, 6585, 19783, 44085, 31184, 2003, 20088, 15969, 42329, 49389, 7501, 55136, 46158, 26141, 37717, 1470, 10599, 25829) % (59722, 45180, 26202, 31878, 13391, 47908, 6431, 38856, 55170, 3022, 16080, 10923, 9173, 37590, 59624, 52820, 7786, 22745, 17906, 55875, 29042, 53787, 15250, 39393, 42659, 43219, 28481, 17127, 3769, 52380, 58137, 51766, 36167, 44338, 62597, 31332, 41678, 38513, 44115) % (57939, 41446, 9208, 58728, 52044, 31536, 26837, 11882, 18906, 52402, 61656, 45740, 44316, 17981, 28294, 13430, 20783, 26797, 17022, 34517, 51572, 7068, 13653, 45214, 1998, 31302, 40026, 6269, 53961, 21899, 11497, 43279, 54380, 31940, 39139, 21988, 36532, 65032, 39749, 50207) % (2735, 35131, 57962, 29571, 22886, 14541, 48163, 54952, 46230, 17651, 39604, 57765, 49943, 31648, 1966, 41206, 29214, 28056, 26165, 52870, 59531, 25902, 53060, 45837, 33754, 9201, 27190, 51309, 20795, 45922, 11591, 42412, 44555, 53415, 59842, 40242, 61532, 6558, 43828, 59576, 14954) % (25818, 20039, 56468, 14423, 15026, 45646, 39033, 5100, 43661, 49819, 57447, 57588, 24168, 48394, 27430, 59992, 42801, 30221, 27081, 55120, 14033, 36294, 31169, 9692, 62632, 47810, 11498, 31124, 44692, 43864, 2301, 44449, 28204, 14049, 3739, 40013, 13374, 60528, 55466, 50886, 3954, 31763) % (51140, 11925, 3691, 9565, 21395, 59730, 40049, 49758, 23768, 8020, 47816, 32259, 36111, 6630, 63557, 27820, 9846, 38700, 12495, 35257, 38101, 45621, 13161, 47630, 46258, 51261, 5021, 42267, 8446, 7478, 21647, 35271, 44696, 2803, 27045, 14746, 42886, 38122, 41953, 49846, 36547, 62499, 35375) % (23654, 14684, 24216, 4066, 59500, 36532, 27442, 28914, 42450, 59208, 48170, 55746, 37745, 50492, 4542, 11605, 13737, 29477, 37455, 15080, 48177, 2427, 43430, 47892, 10895, 18677, 26166, 3948, 8926, 26053, 5974, 41956, 9339, 45808, 59971, 31512, 17323, 15969, 36753, 32130, 58584, 43291, 19320, 22741) % (37362, 28050, 3004, 48390, 52848, 31176, 36293, 9593, 40089, 3375, 15700, 29987, 38849, 5358, 59018, 59557, 20163, 42166, 48680, 22436, 6807, 54805, 10855, 14001, 25813, 9822, 11155, 7259, 17821, 21334, 30833, 3826, 14052, 24360, 42840, 13209, 38195, 10085, 45583, 16258, 14524, 43828, 41450, 48324, 35809) % (60962, 42094, 26827, 12943, 55466, 29462, 62849, 51981, 30249, 24260, 11676, 43729, 50830, 37834, 18918, 8100, 18392, 7323, 8638, 3939, 22861, 15060, 31306, 35389, 19607, 48141, 53303, 48240, 1631, 37101, 26277, 36046, 17657, 58012, 28759, 38291, 54868, 59717, 58712, 30904, 13676, 41953, 12043, 46275, 39706, 60678) % (36442, 25198, 18374, 14663, 23665, 27814, 44896, 13611, 20107, 15516, 38070, 3002, 31251, 13042, 56032, 35299, 9933, 16765, 58451, 62218, 6998, 41653, 5735, 26480, 59941, 63688, 21417, 18543, 37469, 49402, 18901, 23100, 11881, 50963, 41340, 25817, 10994, 43055, 12060, 29750, 41623, 40014, 40497, 52644, 5861, 14801, 3259) % (29154, 45387, 21390, 7373, 45972, 21811, 13839, 39067, 63749, 58507, 27271, 52524, 39142, 33598, 36802, 15637, 49621, 25560, 9461, 43625, 10670, 5093, 46504, 14430, 2875, 44866, 31336, 39841, 42662, 20300, 26979, 57901, 42317, 47512, 16033, 5993, 54318, 62021, 4559, 56520, 46194, 33499, 35175, 57769, 9708, 37500, 10429, 51444) 7
8 Table S.6: A two-level supersaturated design (Lin 1993). Run
9 the number of active factors. Since both D P and D Lin have 14 observations, the number of active factors in this experiment would not be more than 4. However, to be more liberal, we allow to include one more active factor. Thus we considered five cases i = 1,..., 5 with i active factors and for each case, we generated 500 models where the selection of active factors is random and without replacement. The signs of the active factors are also randomly selected from either positive or negative, and the magnitudes are randomly selected from 2 to 10 with replacement. For each model, we generated data 100 times and obtained the True Model Identified Rate (TMIR) and the average true model size, where we recall from Phoa (2014), TMIR is defined as the number of times (over 100 times) that SRRS correctly identifies all active factors and the average true model size is generally the average (over 100 times) of the final model size suggested by SRRS. In our simulations, we followed Phoa (2014) and fixed γ = 1, where γ is the threshold of noise level in SRRS described in the paper. Generally speaking, we choose γ to be approximately 5% 10% of the absolute magnitude of estimate of the first chosen active factor. Readers who are interested in repeating this simulation are encouraged to use the graphical user interface (GUI) package of SRRS that is available in the comprehensive R archive network. Phoa and Lin (2014) describes the implementation of SRRS with several self-explanatory illustrations. Table S.7 compares the simulation results from the two SSDs D P and D Lin. The performance of SRRS is similar using both D P and D Lin as the experimental plans. These distributions seem quite skewed both in TMIR and true model size, which means that the the SRRS performs well no matter if D P or D Lin is used as an experimental plan. Both SSDs are very effective in identifying 1 and 2 active factors in terms of TMIR (at least 96%) and average model sizes, and both are a bit less effective in identifying 3 active factors. The difference between the two SSDs is obvious when SRRS attempts to identify 4 or 5 active factors. When D P is used in cases IV and V, the mean TMIRs and the mean model sizes of SRRS are 4% 7% higher and about 0.1 closer to the true model sizes than the results when D Lin is used. Although the difference of the E(s 2 ) values between the two SSDs is small, there are still some differences in their analysis performance. The SRRS performs better when SSD has a smaller E(s 2 ) value than one with a larger E(s 2 ) value. There are two useful observations in this comparison. First, the above comparison shows that by using the same analysis method (like SRRS) and the same measure (like TMIR or mean model size), a SSD with smaller E(s 2 ) is likely to provide better accuracy. If SSDs are not easily accessible via catalogues or softwares, then SIBSSD becomes a useful and efficient tool for finding SSDs with low E(s 2 ). Second, two SSDs with similar E(s 2 ) values typically result in little changes in the analysis results. For example, in the above comparison, the difference in the SRRS performance between the two SSDs is small and so is their difference in E(s 2 ) values (about 6%). Figure 1 shows if the maximum number of iteration is set at 100, we have a SSD with E(s 2 ) = Is it worth spending the extra amount of time to obtain the E(s 2 )-optimal SSD? Figure 1 shows that the maximum number of iterations is set at step 52, we have a GB SSD with E(s 2 ) = and at step 141, we have a GB SSD with E(s 2 ) = ). This implies that we have to spend almost three times more the amount of time to bring about roughly a 6% improvement in the analysis performance. So our answer is no and this explains why we are satisfied with 9
10 Table S.7: Results from a simulation study to compare abilities of the two designs D P and D Lin for identifying active factors using the Stepwise Response Refinement Screener. Case SSD Min 1st Quartile Median Mean 3rd Quartile Max I TMIR D Lin 97% 99% 100% 99.49% 100% 100% D P 98% 99% 100% 99.31% 100% 100% Size D Lin D P II TMIR D Lin 96% 99% 100% 99.38% 100% 100% D P 97% 99% 100% 99.46% 100% 100% Size D Lin D P III TMIR D Lin 5% 99% 100% 96.96% 100% 100% D P 2% 99% 100% 97.54% 100% 100% Size D Lin D P IV TMIR D Lin 0% 93% 99% 84.33% 100% 100% D P 0% 97% 99% 88.21% 100% 100% Size D Lin D P V TMIR D Lin 0% 20% 86% 64.22% 100% 100% D P 0% 50% 93% 71.06% 99% 100% Size D Lin D P
11 most of our results in our design tables when E r > 90%. References Bulutoglu, D.A., Cheng, C.S. (2004). Construction of E(s 2 )-optimal supersaturated designs. Annals of Statistics 32, Lin, D.K.J. (1993). A new class of supersaturated designs. Technometrics 35, Marley, C.J., Woods, D.C. (2010). A comparison of design and model selection methods for supersaturated experiments. Computational Statistics and Data Analysis 54, Phoa, F.K.H. (2014). The Stepwise Response Refinement Screener (SRRS). Statistica Sinica 24, Phoa, F.K.H. (2014). A graphical user interface platform of the Stepwise Response Refinement Screener for screening experiments. Proceedings of COMPSTAT 2014,
Statistica Sinica Preprint No: SS R2
Statistica Sinica Preprint No: SS-11-291R2 Title The Stepwise Response Refinement Screener (SRRS) Manuscript ID SS-11-291R2 URL http://www.stat.sinica.edu.tw/statistica/ DOI 10.5705/ss.2011.291 Complete
More informationConstruction of optimal Two- Level Supersaturated Designs
RASHI 1 (2) :41-50 (2016) Construction of optimal Two- Level Supersaturated Designs Basudev Kole and Gourav Kumar Rai Department of Statistics, Mathematics & Computer Application Bihar Agricultural University,
More informationAnalysis of Supersaturated Designs via the Dantzig Selector
Analysis of Supersaturated Designs via the Dantzig Selector Frederick K. H. Phoa, Yu-Hui Pan and Hongquan Xu 1 Department of Statistics, University of California, Los Angeles, CA 90095-1554, U.S.A. October
More informationSome Nonregular Designs From the Nordstrom and Robinson Code and Their Statistical Properties
Some Nonregular Designs From the Nordstrom and Robinson Code and Their Statistical Properties HONGQUAN XU Department of Statistics, University of California, Los Angeles, CA 90095-1554, U.S.A. (hqxu@stat.ucla.edu)
More informationStatistica Sinica Preprint No: SS R2
Statistica Sinica Preprint No: SS-2016-0423.R2 Title Construction of Maximin Distance Designs via Level Permutation and Expansion Manuscript ID SS-2016-0423.R2 URL http://www.stat.sinica.edu.tw/statistica/
More informationE(s 2 )-OPTIMAL SUPERSATURATED DESIGNS
Statistica Sinica 7(1997), 929-939 E(s 2 )-OPTIMAL SUPERSATURATED DESIGNS Ching-Shui Cheng University of California, Berkeley Abstract: Tang and Wu (1997) derived a lower bound on the E(s 2 )-value of
More informationMoment Aberration Projection for Nonregular Fractional Factorial Designs
Moment Aberration Projection for Nonregular Fractional Factorial Designs Hongquan Xu Department of Statistics University of California Los Angeles, CA 90095-1554 (hqxu@stat.ucla.edu) Lih-Yuan Deng Department
More informationA Criterion for Constructing Powerful Supersaturated Designs when Effect Directions are Known
A Criterion for Constructing Powerful Supersaturated Designs when Effect Directions are Known Maria L. Weese 1, David J. Edwards 2, and Byran J. Smucker 3 1 Department of Information Systems and Analytics,
More informationAnalysis Methods for Supersaturated Design: Some Comparisons
Journal of Data Science 1(2003), 249-260 Analysis Methods for Supersaturated Design: Some Comparisons Runze Li 1 and Dennis K. J. Lin 2 The Pennsylvania State University Abstract: Supersaturated designs
More informationConstruction and analysis of Es 2 efficient supersaturated designs
Construction and analysis of Es 2 efficient supersaturated designs Yufeng Liu a Shiling Ruan b Angela M. Dean b, a Department of Statistics and Operations Research, Carolina Center for Genome Sciences,
More informationSearching for Powerful Supersaturated Designs
Searching for Powerful Supersaturated Designs MARIA L. WEESE Miami University, Oxford, OH 45056 BYRAN J. SMUCKER Miami University, Oxford, OH 45056 DAVID J. EDWARDS Virginia Commonwealth University, Richmond,
More informationSOME NEW THREE-LEVEL ORTHOGONAL MAIN EFFECTS PLANS ROBUST TO MODEL UNCERTAINTY
Statistica Sinica 14(2004), 1075-1084 SOME NEW THREE-LEVEL ORTHOGONAL MAIN EFFECTS PLANS ROBUST TO MODEL UNCERTAINTY Pi-Wen Tsai, Steven G. Gilmour and Roger Mead National Health Research Institutes, Queen
More informationA Comparison Of Design And. Model Selection Methods. For Supersaturated Experiments
Working Paper M09/20 Methodology A Comparison Of Design And Model Selection Methods For Supersaturated Experiments Christopher J. Marley, David C. Woods Abstract Various design and model selection methods
More informationLooking at a two binary digit sum shows what we need to extend addition to multiple binary digits.
A Full Adder The half-adder is extremely useful until you want to add more that one binary digit quantities. The slow way to develop a two binary digit adders would be to make a truth table and reduce
More informationSearching for D-E cient Equivalent-Estimation Second-Order Split-Plot Designs
Searching for D-E cient Equivalent-Estimation Second-Order Split-Plot Designs NAM-KY NGUYEN VIASM and VNU International School, Hanoi, Vietnam TUNG-DINH PHAM VNU University of Science, Hanoi, Vietnam Several
More informationResearch Article A Novel Differential Evolution Invasive Weed Optimization Algorithm for Solving Nonlinear Equations Systems
Journal of Applied Mathematics Volume 2013, Article ID 757391, 18 pages http://dx.doi.org/10.1155/2013/757391 Research Article A Novel Differential Evolution Invasive Weed Optimization for Solving Nonlinear
More informationAMS 132: Discussion Section 2
Prof. David Draper Department of Applied Mathematics and Statistics University of California, Santa Cruz AMS 132: Discussion Section 2 All computer operations in this course will be described for the Windows
More informationProjection properties of certain three level orthogonal arrays
Metrika (2005) 62: 241 257 DOI 10.1007/s00184-005-0409-9 ORIGINAL ARTICLE H. Evangelaras C. Koukouvinos A. M. Dean C. A. Dingus Projection properties of certain three level orthogonal arrays Springer-Verlag
More informationA RESOLUTION RANK CRITERION FOR SUPERSATURATED DESIGNS
Statistica Sinica 9(1999), 605-610 A RESOLUTION RANK CRITERION FOR SUPERSATURATED DESIGNS Lih-Yuan Deng, Dennis K. J. Lin and Jiannong Wang University of Memphis, Pennsylvania State University and Covance
More informationA MIXED INTEGER QUADRATIC PROGRAMMING MODEL FOR THE LOW AUTOCORRELATION BINARY SEQUENCE PROBLEM. Jozef Kratica
Serdica J. Computing 6 (2012), 385 400 A MIXED INTEGER QUADRATIC PROGRAMMING MODEL FOR THE LOW AUTOCORRELATION BINARY SEQUENCE PROBLEM Jozef Kratica Abstract. In this paper the low autocorrelation binary
More informationUCLA Department of Statistics Papers
UCLA Department of Statistics Papers Title An Algorithm for Constructing Orthogonal and Nearly Orthogonal Arrays with Mixed Levels and Small Runs Permalink https://escholarship.org/uc/item/1tg0s6nq Author
More informationFour Important Number Systems
Four Important Number Systems System Why? Remarks Decimal Base 10: (10 fingers) Most used system Binary Base 2: On/Off systems 3-4 times more digits than decimal Octal Base 8: Shorthand notation for working
More informationA GA Mechanism for Optimizing the Design of attribute-double-sampling-plan
A GA Mechanism for Optimizing the Design of attribute-double-sampling-plan Tao-ming Cheng *, Yen-liang Chen Department of Construction Engineering, Chaoyang University of Technology, Taiwan, R.O.C. Abstract
More informationUCLA Department of Statistics Papers
UCLA Department of Statistics Papers Title The Need of Considering the Interactions in the Analysis of Screening Designs Permalink https://escholarship.org/uc/item/2n50s2qx Authors Frederick K. H. Phoa
More informationDEVELOPMENT IN STANDARD SYSTEM OF METEOROLOGICAL INSTRUMENTS AND METHODS OF OBSERVATION IN CHINA. LU Wenhua, CHEN Xi, LIN Bing, CHONG Wei
DEVELOPMENT IN STANDARD SYSTEM OF METEOROLOGICAL INSTRUMENTS AND METHODS OF OBSERVATION IN CHINA LU Wenhua, CHEN Xi, LIN Bing, CHONG Wei Meteorological Observation Center, China Meteorological Administration,
More informationOptimal Designs for 2 k Experiments with Binary Response
1 / 57 Optimal Designs for 2 k Experiments with Binary Response Dibyen Majumdar Mathematics, Statistics, and Computer Science College of Liberal Arts and Sciences University of Illinois at Chicago Joint
More informationParameter Estimation for the Dirichlet-Multinomial Distribution
Parameter Estimation for the Dirichlet-Multinomial Distribution Amanda Peterson Department of Mathematics and Statistics, University of Maryland, Baltimore County May 20, 2011 Abstract. In the 1998 paper
More informationOptimizing Latin hypercube designs by particle swarm
Stat Comput (2013) 23:663 676 DOI 10.1007/s11222-012-9363-3 Optimizing Latin hypercube designs by particle swarm Ray-Bing Chen Dai-Ni Hsieh Ying Hung Weichung Wang Received: 30 November 2011 / Accepted:
More informationMATH Dr. Halimah Alshehri Dr. Halimah Alshehri
MATH 1101 haalshehri@ksu.edu.sa 1 Introduction To Number Systems First Section: Binary System Second Section: Octal Number System Third Section: Hexadecimal System 2 Binary System 3 Binary System The binary
More informationRandomized Selection on the GPU. Laura Monroe, Joanne Wendelberger, Sarah Michalak Los Alamos National Laboratory
Randomized Selection on the GPU Laura Monroe, Joanne Wendelberger, Sarah Michalak Los Alamos National Laboratory High Performance Graphics 2011 August 6, 2011 Top k Selection on GPU Output the top k keys
More informationSTEAMEST: A Software Tool for Estimation of Physical Properties of Water and Steam
226 JOURNAL OF SOFTWARE, VOL. 4, NO. 3, MAY 2009 STEAMEST: A Software Tool for Estimation of Physical Properties of Water and Steam Muhammad Faheem Department of Chemical Engineering, University of Engineering
More informationA CUDA Solver for Helmholtz Equation
Journal of Computational Information Systems 11: 24 (2015) 7805 7812 Available at http://www.jofcis.com A CUDA Solver for Helmholtz Equation Mingming REN 1,2,, Xiaoguang LIU 1,2, Gang WANG 1,2 1 College
More informationStepsize control for adaptive multiprecision path tracking
Stepsize control for adaptive multiprecision path tracking Daniel J. Bates Jonathan D. Hauenstein Andrew J. Sommese Charles W. Wampler II Abstract Numerical difficulties encountered when following paths
More informationLecture 9 Evolutionary Computation: Genetic algorithms
Lecture 9 Evolutionary Computation: Genetic algorithms Introduction, or can evolution be intelligent? Simulation of natural evolution Genetic algorithms Case study: maintenance scheduling with genetic
More informationAn Algorithm for Constructing Orthogonal and Nearly Orthogonal Arrays with Mixed Levels and Small Runs
An Algorithm for Constructing Orthogonal and Nearly Orthogonal Arrays with Mixed Levels and Small Runs Hongquan Xu Department of Statistics University of California 8130 Math Sciences Bldg Los Angeles,
More informationAlgorithmic Construction of Efficient Fractional Factorial Designs With Large Run Sizes
Algorithmic Construction of Efficient Fractional Factorial Designs With Large Run Sizes Hongquan Xu Department of Statistics University of California Los Angeles, CA 90095-1554 (hqxu@stat.ucla.edu) February
More informationDiscrete Mathematics U. Waterloo ECE 103, Spring 2010 Ashwin Nayak May 17, 2010 Recursion
Discrete Mathematics U. Waterloo ECE 103, Spring 2010 Ashwin Nayak May 17, 2010 Recursion During the past week, we learnt about inductive reasoning, in which we broke down a problem of size n, into one
More informationUSING REGULAR FRACTIONS OF TWO-LEVEL DESIGNS TO FIND BASELINE DESIGNS
Statistica Sinica 26 (2016, 745-759 doi:http://dx.doi.org/10.5705/ss.202014.0099 USING REGULAR FRACTIONS OF TWO-LEVEL DESIGNS TO FIND BASELINE DESIGNS Arden Miller and Boxin Tang University of Auckland
More informationCRYPTOGRAPHIC COMPUTING
CRYPTOGRAPHIC COMPUTING ON GPU Chen Mou Cheng Dept. Electrical Engineering g National Taiwan University January 16, 2009 COLLABORATORS Daniel Bernstein, UIC, USA Tien Ren Chen, Army Tanja Lange, TU Eindhoven,
More informationHIDDEN PROJECTION PROPERTIES OF SOME NONREGULAR FRACTIONAL FACTORIAL DESIGNS AND THEIR APPLICATIONS 1
The Annals of Statistics 2003, Vol. 31, No. 3, 1012 1026 Institute of Mathematical Statistics, 2003 HIDDEN PROJECTION PROPERTIES OF SOME NONREGULAR FRACTIONAL FACTORIAL DESIGNS AND THEIR APPLICATIONS 1
More informationDepartment of Computer Science University at Albany, State University of New York Solutions to Sample Discrete Mathematics Examination II (Fall 2007)
Department of Computer Science University at Albany, State University of New York Solutions to Sample Discrete Mathematics Examination II (Fall 2007) Problem 1: Specify two different predicates P (x) and
More informationConversions between Decimal and Binary
Conversions between Decimal and Binary Binary to Decimal Technique - use the definition of a number in a positional number system with base 2 - evaluate the definition formula ( the formula ) using decimal
More informationOn the Computational Complexity of the Discrete Pascal Transform
6 th International Conference Logic and Applications LAP 207, September 8-22, 207, Dubrovnik, Croatia On the Computational Complexity of the Discrete Pascal Transform Dušan B. Gajić, Radomir S. Stanković
More informationWAITING-TIME DISTRIBUTION FOR THE r th OCCURRENCE OF A COMPOUND PATTERN IN HIGHER-ORDER MARKOVIAN SEQUENCES
WAITING-TIME DISTRIBUTION FOR THE r th OCCURRENCE OF A COMPOUND PATTERN IN HIGHER-ORDER MARKOVIAN SEQUENCES Donald E. K. Martin 1 and John A. D. Aston 2 1 Mathematics Department, Howard University, Washington,
More informationUsing Animal Instincts to Design Efficient Biomedical Studies
Using Animal Instincts to Design Efficient Biomedical Studies Jiaheng Qiu a, Ray-Bing Chen b, Weichung Wang c, Weng Kee Wong d a Department of Biostatistics, University of California, Los Angeles, CA 90095,
More informationDE [39] PSO [35] ABC [7] AO k/maxiter e-20
3. Experimental results A comprehensive set of benchmark functions [18, 33, 34, 35, 36, 37, 38] has been used to test the performance of the proposed algorithm. The Appendix A (Table A1) presents the functions
More informationIndividual Household Financial Planning
Individual Household Financial Planning Igor Osmolovskiy Cambridge Systems Associates Co- workers: Michael Dempster, Elena Medova, Philipp Ustinov 2 ialm helps households to ialm individual Asset Liability
More informationOptimal Selection of Blocked Two-Level. Fractional Factorial Designs
Applied Mathematical Sciences, Vol. 1, 2007, no. 22, 1069-1082 Optimal Selection of Blocked Two-Level Fractional Factorial Designs Weiming Ke Department of Mathematics and Statistics South Dakota State
More informationAn efficient ADMM algorithm for high dimensional precision matrix estimation via penalized quadratic loss
An efficient ADMM algorithm for high dimensional precision matrix estimation via penalized quadratic loss arxiv:1811.04545v1 [stat.co] 12 Nov 2018 Cheng Wang School of Mathematical Sciences, Shanghai Jiao
More informationSuperiorized Inversion of the Radon Transform
Superiorized Inversion of the Radon Transform Gabor T. Herman Graduate Center, City University of New York March 28, 2017 The Radon Transform in 2D For a function f of two real variables, a real number
More informationShortest Lattice Vector Enumeration on Graphics Cards
Shortest Lattice Vector Enumeration on Graphics Cards Jens Hermans 1 Michael Schneider 2 Fréderik Vercauteren 1 Johannes Buchmann 2 Bart Preneel 1 1 K.U.Leuven 2 TU Darmstadt SHARCS - 10 September 2009
More informationA CARTOGRAPHIC DATA MODEL FOR BETTER GEOGRAPHICAL VISUALIZATION BASED ON KNOWLEDGE
A CARTOGRAPHIC DATA MODEL FOR BETTER GEOGRAPHICAL VISUALIZATION BASED ON KNOWLEDGE Yang MEI a, *, Lin LI a a School Of Resource And Environmental Science, Wuhan University,129 Luoyu Road, Wuhan 430079,
More informationTime Aggregation for Network Design to Meet Time-Constrained Demand
20th International Congress on Modelling and Simulation, Adelaide, Australia, 1 6 December 2013 www.mssanz.org.au/modsim2013 Time Aggregation for Network Design to Meet Time-Constrained Demand N. Boland
More informationGENERALIZED ARYABHATA REMAINDER THEOREM
International Journal of Innovative Computing, Information and Control ICIC International c 2010 ISSN 1349-4198 Volume 6, Number 4, April 2010 pp. 1865 1871 GENERALIZED ARYABHATA REMAINDER THEOREM Chin-Chen
More informationFree Search in Multidimensional Space
Free Search in Multidimensional Space Technology School, Maritime and Technology Faculty Southampton Solent University Content Aims Test Problems Bump test function Michalewicz test function Norwegian
More informationA Recursive Formula for the Kaplan-Meier Estimator with Mean Constraints
Noname manuscript No. (will be inserted by the editor) A Recursive Formula for the Kaplan-Meier Estimator with Mean Constraints Mai Zhou Yifan Yang Received: date / Accepted: date Abstract In this note
More informationModel Predictive Controller of Boost Converter with RLE Load
Model Predictive Controller of Boost Converter with RLE Load N. Murali K.V.Shriram S.Muthukumar Nizwa College of Vellore Institute of Nizwa College of Technology Technology University Technology Ministry
More informationTwo problems to be solved. Example Use of SITATION. Here is the main menu. First step. Now. To load the data.
Two problems to be solved Example Use of SITATION Mark S. Daskin Department of IE/MS Northwestern U. Evanston, IL 1. Minimize the demand weighted total distance (or average distance) Using 10 facilities
More informationFactorisation of RSA-704 with CADO-NFS
Factorisation of RSA-704 with CADO-NFS Shi Bai, Emmanuel Thomé, Paul Zimmermann To cite this version: Shi Bai, Emmanuel Thomé, Paul Zimmermann. Factorisation of RSA-704 with CADO-NFS. 2012. HAL Id: hal-00760322
More informationWhat Every Programmer Should Know About Floating-Point Arithmetic DRAFT. Last updated: November 3, Abstract
What Every Programmer Should Know About Floating-Point Arithmetic Last updated: November 3, 2014 Abstract The article provides simple answers to the common recurring questions of novice programmers about
More informationPC ARC/INFO and Data Automation Kit GIS Tools for Your PC
ESRI PC ARC/INFO and Data Automation Kit GIS Tools for Your PC PC ARC/INFO High-quality digitizing and data entry Powerful topology building Cartographic design and query Spatial database query and analysis
More informationOn High-Dimensional Cross-Validation
On High-Dimensional Cross-Validation BY WEI-CHENG HSIAO Institute of Statistical Science, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 11529, Taiwan hsiaowc@stat.sinica.edu.tw 5 WEI-YING
More informationCSC 4510 Machine Learning
10: Gene(c Algorithms CSC 4510 Machine Learning Dr. Mary Angela Papalaskari Department of CompuBng Sciences Villanova University Course website: www.csc.villanova.edu/~map/4510/ Slides of this presenta(on
More informationA GENERAL CONSTRUCTION FOR SPACE-FILLING LATIN HYPERCUBES
Statistica Sinica 6 (016), 675-690 doi:http://dx.doi.org/10.5705/ss.0015.0019 A GENERAL CONSTRUCTION FOR SPACE-FILLING LATIN HYPERCUBES C. Devon Lin and L. Kang Queen s University and Illinois Institute
More informationBest subset selection via bi-objective mixed integer linear programming
Best subset selection via bi-objective mixed integer linear programming Hadi Charkhgard a,, Ali Eshragh b a Department of Industrial and Management Systems Engineering, University of South Florida, Tampa,
More informationAFT. Advanced Friction Tester. Static and dynamic coefficient of friction Fast, repeatable measurements Compliant to multiple standards
AFT Advanced Friction Tester Static and dynamic coefficient of friction Fast, repeatable measurements Compliant to multiple standards AFT Advanced Friction Tester VERSATILE INSTRUMENT REPEATABLE MEASUREMENTS
More informationLecture 11. Advanced Dividers
Lecture 11 Advanced Dividers Required Reading Behrooz Parhami, Computer Arithmetic: Algorithms and Hardware Design Chapter 15 Variation in Dividers 15.3, Combinational and Array Dividers Chapter 16, Division
More informationTHE EIGENVALUE PROBLEM
THE EIGENVALUE PROBLEM Let A be an n n square matrix. If there is a number λ and a column vector v 0 for which Av = λv then we say λ is an eigenvalue of A and v is an associated eigenvector. Note that
More informationRandom Number Generation Is Getting Harder It s Time to Pay Attention
SESSION ID: PDAC-F03 Random Number Generation Is Getting Harder It s Time to Pay Attention Richard Moulds General Manager Whitewood Richard Hughes Laboratory Fellow (Retired) Los Alamos National Laboratory
More informationThe statistics used in this report have been compiled before the completion of any Post Results Services.
Course Report 2015 Subject Mathematics Level National 5 The statistics used in this report have been compiled before the completion of any Post Results Services. This report provides information on the
More informationIncremental Latin Hypercube Sampling
Incremental Latin Hypercube Sampling for Lifetime Stochastic Behavioral Modeling of Analog Circuits Yen-Lung Chen +, Wei Wu *, Chien-Nan Jimmy Liu + and Lei He * EE Dept., National Central University,
More informationP8130: Biostatistical Methods I
P8130: Biostatistical Methods I Lecture 2: Descriptive Statistics Cody Chiuzan, PhD Department of Biostatistics Mailman School of Public Health (MSPH) Lecture 1: Recap Intro to Biostatistics Types of Data
More informationAppendix A.0: Approximating other performance measures
Appendix A.0: Approximating other performance measures Alternative definition of service level and approximation. The waiting time is defined as the minimum of virtual waiting time and patience. Define
More informationPairings at High Security Levels
Pairings at High Security Levels Michael Naehrig Eindhoven University of Technology michael@cryptojedi.org DoE CRYPTODOC Darmstadt, 21 November 2011 Pairings are efficient!... even at high security levels.
More informationMULTIPLE-OBJECTIVE DESIGNS IN A DOSE-RESPONSE EXPERIMENT
New Developments and Applications in Experimental Design IMS Lecture Notes - Monograph Series (1998) Volume 34 MULTIPLE-OBJECTIVE DESIGNS IN A DOSE-RESPONSE EXPERIMENT BY WEI ZHU AND WENG KEE WONG 1 State
More informationSparse solver 64 bit and out-of-core addition
Sparse solver 64 bit and out-of-core addition Prepared By: Richard Link Brian Yuen Martec Limited 1888 Brunswick Street, Suite 400 Halifax, Nova Scotia B3J 3J8 PWGSC Contract Number: W7707-145679 Contract
More informationCharacterizations of indicator functions of fractional factorial designs
Characterizations of indicator functions of fractional factorial designs arxiv:1810.08417v2 [math.st] 26 Oct 2018 Satoshi Aoki Abstract A polynomial indicator function of designs is first introduced by
More informationPerformance Metrics for Computer Systems. CASS 2018 Lavanya Ramapantulu
Performance Metrics for Computer Systems CASS 2018 Lavanya Ramapantulu Eight Great Ideas in Computer Architecture Design for Moore s Law Use abstraction to simplify design Make the common case fast Performance
More informationAn Experimental Evaluation of Passage-Based Process Discovery
An Experimental Evaluation of Passage-Based Process Discovery H.M.W. Verbeek and W.M.P. van der Aalst Technische Universiteit Eindhoven Department of Mathematics and Computer Science P.O. Box 513, 5600
More informationChapter 1 Statistical Inference
Chapter 1 Statistical Inference causal inference To infer causality, you need a randomized experiment (or a huge observational study and lots of outside information). inference to populations Generalizations
More informationResampling Methods CAPT David Ruth, USN
Resampling Methods CAPT David Ruth, USN Mathematics Department, United States Naval Academy Science of Test Workshop 05 April 2017 Outline Overview of resampling methods Bootstrapping Cross-validation
More informationPrimality Testing- Is Randomization worth Practicing?
Primality Testing- Is Randomization worth Practicing? Shubham Sahai Srivastava Indian Institute of Technology, Kanpur ssahai@cse.iitk.ac.in April 5, 2014 Shubham Sahai Srivastava (IITK) Primality Test
More informationOPTIMAL DESIGNS FOR 2 k FACTORIAL EXPERIMENTS WITH BINARY RESPONSE
Statistica Sinica 26 (2016), 385-411 doi:http://dx.doi.org/10.5705/ss.2013.265 OPTIMAL DESIGNS FOR 2 k FACTORIAL EXPERIMENTS WITH BINARY RESPONSE Jie Yang 1, Abhyuday Mandal 2 and Dibyen Majumdar 1 1 University
More informationk-protected VERTICES IN BINARY SEARCH TREES
k-protected VERTICES IN BINARY SEARCH TREES MIKLÓS BÓNA Abstract. We show that for every k, the probability that a randomly selected vertex of a random binary search tree on n nodes is at distance k from
More informationCHAPTER 2 EXTRACTION OF THE QUADRATICS FROM REAL ALGEBRAIC POLYNOMIAL
24 CHAPTER 2 EXTRACTION OF THE QUADRATICS FROM REAL ALGEBRAIC POLYNOMIAL 2.1 INTRODUCTION Polynomial factorization is a mathematical problem, which is often encountered in applied sciences and many of
More informationLandslide Classification: An Object-Based Approach Bryan Zhou Geog 342: Final Project
Landslide Classification: An Object-Based Approach Bryan Zhou Geog 342: Final Project Introduction One type of natural hazard that people are familiar with is landslide. Landslide is a laymen term use
More informationBranch Prediction based attacks using Hardware performance Counters IIT Kharagpur
Branch Prediction based attacks using Hardware performance Counters IIT Kharagpur March 19, 2018 Modular Exponentiation Public key Cryptography March 19, 2018 Branch Prediction Attacks 2 / 54 Modular Exponentiation
More informationFault Tolerance Technique in Huffman Coding applies to Baseline JPEG
Fault Tolerance Technique in Huffman Coding applies to Baseline JPEG Cung Nguyen and Robert G. Redinbo Department of Electrical and Computer Engineering University of California, Davis, CA email: cunguyen,
More informationDesign and FPGA Implementation of Radix-10 Algorithm for Division with Limited Precision Primitives
Design and FPGA Implementation of Radix-10 Algorithm for Division with Limited Precision Primitives Miloš D. Ercegovac Computer Science Department Univ. of California at Los Angeles California Robert McIlhenny
More informationA Blockwise Descent Algorithm for Group-penalized Multiresponse and Multinomial Regression
A Blockwise Descent Algorithm for Group-penalized Multiresponse and Multinomial Regression Noah Simon Jerome Friedman Trevor Hastie November 5, 013 Abstract In this paper we purpose a blockwise descent
More informationMarkscheme May 2016 Mathematical studies Standard level Paper 1
M16/5/MATSD/SP1/ENG/TZ1/XX/M Markscheme May 016 Mathematical studies Standard level Paper 1 4 pages M16/5/MATSD/SP1/ENG/TZ1/XX/M This markscheme is the property of the International Baccalaureate and must
More informationCryptographic D-morphic Analysis and Fast Implementations of Composited De Bruijn Sequences
Cryptographic D-morphic Analysis and Fast Implementations of Composited De Bruijn Sequences Kalikinkar Mandal, and Guang Gong Department of Electrical and Computer Engineering University of Waterloo Waterloo,
More informationINTELLIGENT SEARCH FOR AND MINIMUM ABERRATION DESIGNS
Statistica Sinica 8(1998), 1265-1270 INTELLIGENT SEARCH FOR 2 13 6 AND 2 14 7 MINIMUM ABERRATION DESIGNS Jiahua Chen University of Waterloo Abstract: Among all 2 n k regular fractional factorial designs,
More informationImage Compression. 1. Introduction. Greg Ames Dec 07, 2002
Image Compression Greg Ames Dec 07, 2002 Abstract Digital images require large amounts of memory to store and, when retrieved from the internet, can take a considerable amount of time to download. The
More informationLecture 11. Linear Soft Margin Support Vector Machines
CS142: Machine Learning Spring 2017 Lecture 11 Instructor: Pedro Felzenszwalb Scribes: Dan Xiang, Tyler Dae Devlin Linear Soft Margin Support Vector Machines We continue our discussion of linear soft margin
More informationBright Advance Corporation
USER INSTRUCTIONS TABLE OF CONTENTS INSTRUCTIONS FOR USE2 PREPARING TO USE THE SCALE2 DISPLAYS3 KEYBOARD FUNCTION4 OPERATION7 COUNTING14 DIFFERENT KEYBOARD TYPES21 INTERFACE31 POWER SOURCES40 1 INSTRUCTIONS
More informationRank parameters for Bland Altman plots
Rank parameters for Bland Altman plots Roger B. Newson May 2, 8 Introduction Bland Altman plots were introduced by Altman and Bland (983)[] and popularized by Bland and Altman (986)[2]. Given N bivariate
More information: TEACHING DIFFERENTIAL EQUATIONS WITH AN ENGINEERING FOCUS
2006-915: TEACHING DIFFERENTIAL EQUATIONS WITH AN ENGINEERING FOCUS Stephen Pennell, University of Massachusetts-Lowell Stephen Pennell is a Professor in the Department of Mathematical Sciences at the
More informationM 225 Test 1 B Name SHOW YOUR WORK FOR FULL CREDIT! Problem Max. Points Your Points Total 75
M 225 Test 1 B Name SHOW YOUR WORK FOR FULL CREDIT! Problem Max. Points Your Points 1-13 13 14 3 15 8 16 4 17 10 18 9 19 7 20 3 21 16 22 2 Total 75 1 Multiple choice questions (1 point each) 1. Look at
More informationUtilization and Provision of Geographical Name Information on the Basic Map of Japan*
UNITED NATIONS WORKING PAPER GROUP OF EXPERTS NO. 1/9 ON GEOGRAPHICAL NAMES Twenty-eight session 28 April 2 May 2014 Item 9 of the Provisional Agenda Activities relating to the Working Group on Toponymic
More information