1-1 定義
- upper triangular matrix : 矩陣對角線下方都為零 ->strictly upper triangular : 連對角線都為零
- lower triangular matrix : 矩陣對角線上方都為零->strictly lower triangular : 連對角線都為零
- 零矩陣
- I矩陣
- square matrix : m(列)=n(欄),為方矩陣
- diagonal matrix : 只有對角線有數字
1-2 運算
- 加減法 : 矩陣size必須一樣
if and only if 若且唯若 (iff) : 在數學定義必為雙向都要成立,左邊對,右邊也要對,為等價關係。
- 純量積 : 常數倍數乘上矩陣
- 乘法 : A = m*‘n’ , B = ‘n’*p ,A,B才能相乘
Property :
1. A = m*n
A・In = A = Im・A (A不是方陣仍正確)
2. (AB)C = A(BC) 結合律
3. A(B+C) = AB+AC 分配率
(A+B)C = AC+BC
4. (cA)B = A(cB) 純量積
不恆成立:
1. AB = BA 未必成立
2. A^n = 0 -> A = 0 未必成立
3. A^2 = A -> A=0 or I 未必成立
4. A=\=0 , B =\= 0 , 未必AB =\= 0
5. AB = AC , A=\=0 , 未必 B = C
6. x^n = A 未必至多 n 個解
7. (A+B)^2 = A^2 + 2AB + B^2 未必成立
-> = (A+B)(A+B) = A^2+ AB + BA + B^2
( I+A )^2 = I^2 + 2A + A^2 ( I可以進行交換 )
*證明上三角相乘仍為上三角(看備忘錄)
- 矩陣分割 (方塊矩陣) :5*5 -> 2*2的方塊矩陣,要滿足外面可相乘,裡面也可相乘
四個工具:
行展:(主流) column vector
AX = X1a1+ X2a2+ ...+ Xnan = 為A的行向量的線性組合linear combination(取倍數再相加 e1 ,e2 ,e3之類)
*X是向量係數coefficients列展:
XA = X1a1+ X2a2+ ...+ Xmam行切:(主流)
A = m*n ; B= n*p
AB = A [ b1 , b2 , ... , bp ] = [ Ab1, Ab2, ... , Abp ]列切:
A = m*n ; B= n*p
AB = [ a1 , a2 , ... am ]^-1 B = [ a1B , a2B , ... , amB ]
定義:
- A^T 轉置矩陣( transpose )
- A^H 共軛轉置矩陣 ( conjugate transpose ) -> transpose 過後再取bar (共軛係數加負號)
Property :
1. (A+-B)^T = A^T + B^T
2. ( cA )^T = c * A^T
3. ( cA + dB )^T = cA^T + dB^T
4. ( A^T )^T = A
5. ( AB )^T = B^T*A^T ; ( ABC )^T = C^T*B^T*A^T
6. ( A+-B )^H = A^H +- B^H
7. ( cA ) ^H = c^bar * A^H
8. ( cA +- dB )^H = c^barA^H +- d^barB^H
9. ( A^H )^H = A
10. ( AB )^H = B^H*A^H
定義:
A = m*n- A^T = A : 稱A為 sysmetric matrix
- A^T = -A : 稱A為 skew-symmetric (斜對稱)
- A^H = A : 稱A為 Hermitian
- A^H = -A : 稱A為skew-Hermitian
*次方和轉置可以交換 : ( A^K )^T = ( A^T ) ^K
定義:
A : n x n 定義 tr (A) = a11 +a22 + a33... +ann = 對角項數值相加,稱之為A的trace。
Property :
1. tr(A +- B ) = tr(A) +- tr(B)
2. tr(cA) = c * tr(A)
3. tr(cA +- dB ) = c*tr(A) +- d*tr(B)
4. tr(In) = n
5. tr(A^T) = tr(A)
6. tr(A^H) = (tr(A))^bar = tr(A^bar)
7. tr(AB) = tr(BA)