可逆矩阵求特征值，其特征值矩阵得到的为嘛不正交啦？？？

小周周

两个矩阵，分别求特征值和特征矩阵，按理说特征矩阵应该是正交的，满足T的转置等于T的逆，但是我求出来的却不是！求问原因。Yg=[0.302794262821957 - 3.79296757057053i,0.00000000000000 + 0.00000000000000i,0.0874945715832985 + 0.641221324267202i;0.00000000000000 + 0.00000000000000i,0.00000000000000 - 5.48550000000000i,0.00000000000000 + 0.00000000000000i;0.0874945715832984 + 0.641221324267202i,0.00000000000000 + 0.00000000000000i,170.000000000000 + 0.00000000000000i]
3 a- w; u) F6 Z- ~3 a

' e9 G9 _  K  Y3 o$ d7 M! RYg2=[0.972678496874197 - 2.19149951986028i
0.247515468245944 + 0.721616638839712i
0.289597088366804 + 1.08389127598852i;
9 D4 n; ^: w% r3 U0.247515468245944 + 0.721616638839712i
3 e% `3 V" O- ]0 {5 C/ Z) {1 S0.270106378209774 - 1.95560247734671i
0.196726501279043 + 1.22862731973151i;
; z5 D9 R. r6 O7 o* y0.289597088366804 + 1.08389127598852i
0.196726501279043 + 1.22862731973151i
% W( Y+ S. l; H( C0.163424123781682 - 2.34585747664809i];


/ Q& S$ x  T9 t# m$ X' e' ?[T1,X1]=eig(Yg);[T2,X2]=eig(Yg2);
T1*T1'不等于E，应该不是正交矩阵啦？

小周周

已经知道原因了。复矩阵的正交要取hermite 就是要让虚部取个负号再转置

cxxgtxy

回复 2# 小周周
3 m* a" o( o& W5 y
* y/ m, F5 z& `
    这个matlab help里面写的非常清楚，你help一下就知道了