Table 1. The algorithm of OCMA calculates the right outcome. Eigenvectors (or singular vectors) are compared using Infinite norm $({\Vert \text{*}\Vert }_{\mathrm{\infty }}=\text{max}\{\left|x_1\right|,...,\left|x_n\right|\left\}\right)$, Taxicab norm $({\Vert \text{*}\Vert }_1={\displaystyle \sum }_{i=1}^n|x_i\left|\right)$, and Euclidean norm $({\Vert \text{*}\Vert }_2=\sqrt{{\displaystyle \sum }_{i=1}^n{x}_i{}^2})$. We use ${\lambda }_{ocma}$ and $S_{ocma}$ to denote the vectors calculated by OCMA and ${\lambda }_{matlab}$ and $S_{matlab}$ to denote the vectors calculated by MATLAB (version 2012b). The comparison for eigen-decomposition is presented in the upper table, and the comparison for singular-value decomposition is presented in the lower one. In the upper table, N denotes the number of individuals. In lower table, N = 1,000,000, and M denotes the number of genetic markers. The configuration of the personal computer: Intel Core i7-6700 CPU (4 cores), Memory = 24GB. Disk = Samsung SSD 850 EVO 250GB. The operating system is Windows 7. Time is measured by wall-clock (instead of CPU time).

N	$\Vert {\lambda }_{ocma}-{\lambda }_{matlab}\Vert /\Vert {\lambda }_{matlab}\Vert$			Computation time (s)
	${\Vert \text{*}\Vert }_{\mathrm{\infty }}$	${\Vert \text{*}\Vert }_1$	${\Vert \text{*}\Vert }_2$	OCMA	MATLAB
1000	5.2*10−7	7.6*10−8	1.1*10−7	0.1	0.4
2000	2.8*10−7	8.1*10−8	1.0*10−7	0.5	2.8
5000	4.3*10−7	2.1*10−7	2.4*10−7	6.4	35.5
10000	1.6*10−6	5.7*10−7	6.3*10−7	46.2	294.0
20000	3.3*10−6	4.4*10−6	4.3*10−6	246.4	2905.1

M	$\Vert s_{ocma}-s_{matlab}\Vert /\Vert s_{matlab}\Vert$			Computation time (s)
	${\Vert \text{*}\Vert }_{\mathrm{\infty }}$	${\Vert \text{*}\Vert }_1$	${\Vert \text{*}\Vert }_2$	OCMA	MATLAB
100	2.9*10−5	1.2*10−4	8.9*10−5	10.2	12.3
200	2.1*10−5	9.2*10−5	6.6*10−5	14.8	29.6
500	1.9*10−5	7.7*10−5	5.6*10−5	47.5	90.4
1000	1.6*10−5	1.3*10−4	8.5*10−5	89.5	206.4
2000	1.2*10−5	1.1*10−4	7.1*10−5	226.1	726.9