TABLE 3

Error counts and optimization summaries for Example 1

Estimated configurationMethodOptimized criterionIteration
(13, 2, 2, 135, 91)S(C) T = 301,523,081734,507
(12, 2, 2, 135, 101)S(C)SA T = 301,522,899308,452
1,522,899469,098
1,522,899532,671
S(C) T = 101,522,899533,039
(13, 29, 3, 1134, 101)S(C) T = 31,522,864495,801
S(C)SA T0 = 301,522,864734,234
(13, 27, 4, 790, 125)S(C)SA T0 = 301,519,256680,734
True configuration1,522,637
(17, 108, 7, 4130, 168)ML SA T0 = 30–61.1957,193
(24, 294, 14, 15701, 139)ML–415.6226,453
(27, 372, 18, 19022, 96)ML–478.6921,908
(28, 375, 17, 19062, 4)ML SA T0 = 1–491.3273,616
(25, 323, 15, 17470, 110)ML–509.1762,518
(12, 22, 99, 197) at trueML started–27.8362
(12, 0, 0, 0, 0)True configuration–89.0

Optimized criterion is maximized value of log-likelihood or score iteration (iteration at which optimum was first encountered). Configuration is (no. f.s., no. moves, no. group moves, E1, E2), where no. f.s. is number of full-sib groups; no. moves is number of moves to correct configuration; no. g.m. is number of group moves; E1 is number of full-sib pairs misclassified as unrelated; E2 is number of unrelated pairs misclassified as full-sibs; ML is the maximum-likelihood method; S(C) is the pairwise scoring method; and SA T is simulated annealing with starting temperature T.