Supporting Tables - Design of Experiments for Reinforcement Learning - page 175

Table C.3 Convergent subregion ranges for the TBU problem.
Subregion # Hidden nodes ʱ mag ʱ ratio ʻ ʳ

10 − 3 ,8 . 398 10 − 3 ) 10 − 2 ,0 . 298 ) 12 [26, 61) [6 . 007 × × [1.943, 8.000) [7 . 075 × [0.957, 0.988) [0.974, 1.000)

25 [26, 61)

[ 6 . 589

× 10 − 3 ,8 . 398 × 10 − 3 ) [1.943, 3.497)

[ 7 . 075

× 10 − 2 ,0 . 193 ) [0.961, 0.988) [0.887, 0.965)

10 − 3 ,0 . 010 ) 10 − 2 ) 31 [40, 61) [8 . 398 × [2.718, 8.000) [0 . 000, 6 . 188 × [0.957, 0.985) [0.888, 0.949)

32 [30, 61) [8 . 398 × 10 − 3 ,0 . 010 ) [2.718, 8.000) [7 . 115 × 10 − 2 ,0 . 166 ) [0.956, 0.985) [0.888, 0.949)

10 − 3 ,0 . 010 )

35 [30, 59) [8 . 398 × [1.000, 7.619) [0.000, 0.229) [0.957, 0.979) [0.949, 1.000)

10 − 3 ,9 . 154 10 − 3 ) 63 [30, 61) [8 . 842 × × [4.484, 5.770) [0.000, 0.229) [0.979, 0.989) [0.949, 0.997)

[9 . 451 × 10 − 3 ,0 . 010 ) 70 [33, 61) [1.000, 8.000) [0.000, 0.229) [0.979, 0.989) [0.971, 0.994)

Table C.4 Kriging model parameters for the TBU problem.
Trend parameters Covariance parameters

Subregion # Intercept Hidden nodes ʱ mag ʱ mag ʻ ʳ Hidden nodes ʱ mag ʱ mag ʻ ʳ

12 50.958 0.932 − 0.313 − 0.289 0.486 11.832 − 0.191 0.394 0.73 0.503 2.000 0.136 0.797

25 56.278 − 0.31 − 0.685 0.392 − 0.431 4.611 0.435 0.000 2.000 2.000 2.000 0.098 0.403

31 54.508 0.176 − 0.291 0.766 − 0.626 6.891 − 0.060 2.000 2.000 0.393 2.000 0.133 2.000

32 52.653 − 0.117 0.671 0.653 0.054 8.255 − 0.264 0.722 2.000 0.684 2.000 0.193 0.694

35 53.721 0.313 − 0.033 − 2.177 − 0.197 8.716 − 0.990 2.000 2.000 0.386 2.000 0.194 0.374

− − − −

63 60.691 0.885 0.388 0.048 1.062 2.31 1.106 0.046 2.000 2.000 0.449 0.292 0.570

70 58.797 − 0.293 0.347 2.334 − 1.913 2.554 1.418 0.577 0.659 0.874 1.929 0.000 0.376

Table C.5 Convergent subregion ranges for the TTBU problem.
Subregion # ʱ mag ʱ ratio ʻ ʳ

24 [0.004, 0.008) [2.000, 2.592) [0.493, 0.700) [0.976, 0.990) [0.850, 0.970)