Embedded Files
7. 正交化
7. 正交化
7. Orthogonalization
7. Orthogonalization
非正交设计可以被正交化，图 1 中的矩阵是一个非正交矩阵，正交化后的结果见下图
Non-orthogonal matrices can be orthogonalized. The matrix in the figure 1 is a non-orthogonal matrix. The result of orthogonalization is shown as following
The orthogonalization result for example 1
The orthogonalization result for example 1
我们看到了，第四列(x3)正交化后变成了 0 向量。证明前四列是一个完全相关组。从这一列起，后面的列全都不正常了，这个矩阵整体上不能作为试验设计用。
请注意，上图中正交性校验矩阵显示的数据是不准确的，有些数据被隐藏了，这可以调节这个网格的宽度，看到真实的数据，如下图所示。
Check orthogonality, ρ is the residual rate of the corresponding vector, and V is the orthogonalization volume.
Check orthogonality, ρ is the residual rate of the corresponding vector, and V is the orthogonalization volume.
We can see that the fourth column (x3) becomes a 0 vector after orthogonalization. It prove that the first four columns are a completely correlated group. This matrix as a whole cannot be used for experimental design.
Please note that the data displayed by the orthogonality check matrix in the above figure is inaccurate. Some data is hidden. This can adjust the width of this grid to see the real data, as shown in the left figure .
8. 性能点绘
8. 性能点绘
上述参数可以点绘在坐标上，下图是我们构造的弱相关矩阵 w21h7o 的性能图，点绘规律整齐。
8 . Performance plot
8 . Performance plot
The above parameters can be drawn on the coordinates, the weak correlation matrix w21h7o was constructed by me, and its dots are neatly drawn.
mcc
P-max
ρ
V
例 1 的性能点绘则十分凌乱。
The performance plot of Example 1 is very messy.
mcc
P-max
ρ
There are many points invisible here
V
Page updated
Google Sites
Report abuse