DR12 - Nearly Orthogonal Latin Hypercubes - 17 to 22 dimensions

Nearly orthogonal Latin hypercubes of order 7 optimized using an evolutionary algorithm to minimize the modified L2 discrepancy (ML2), maximize the euclidean maximin distance (EMm), minimize the maximum pairwise correlation (MPwC), and minimize the condition number (Cond).

Properties

Order	Dimensions	Number of Points	ML2	EMm	MPwC	Cond
7	17	129	3.2348	1.8665	0.00693	0.0
	18	129	5.2220	2.1461	0.00693	0.0
	19	129	8.3445	2.2445	0.00693	0.0
	20	129	13.263	2.3667	0.00693	0.0
	21	129	21.303	2.41141	0.00693	0.0
	22	129	33.464	2.5630	0.00693	1.0413

Columns removed for the 17 dimensions design are {8, 11, 12, 14, 17}, for the 18 dimensions design {8, 11, 12, 17}, for the 19 dimensions design {10, 15, 22}, for the 20 dimensions design {8, 12}, and for the 21 dimensions design {15}.

Dowloads

Points sets should be read row major; an entire row is a point and each column is one dimension for a point.

Cite As

F.-M. De Rainville, C. Gagné, O. Teytaud, and D. Laurendeau. Evolutionary optimization of low-discrepancy sequences. ACM Trans. Model. Comput. Simul., 22(2):9:1–9:25, 2012.
Original Version
LVSN Version