[선형대수] Lecture 20: Cramer's rule, inverse matrix, and volume

이재호·2025년 3월 14일

선형대수

목록 보기
19/31

https://ocw.mit.edu/courses/18-06-linear-algebra-spring-2010/video_galleries/video-lectures/

이번 강의에서는 다음과 같은 내용을 다룬다.

1. Formula for A1\text{1. Formula for $A^{-1}$}
2. Cramer’s Rule for X=A1b\text{2. Cramer's Rule for $X=A^{-1}b$}
3. detA = Volume of box\text{3. $|det A|$ = Volume of box}

먼저 1. formular for A1A^{-1}을 알아보자.
2 x 2 matrix의 역함수 공식은 다음과 같다.

[abcd]1=1adbc[dbca]\begin{bmatrix} a & b \\ c & d \end{bmatrix}^{-1} = \frac{1}{ad-bc} \begin{bmatrix} d & -b \\ -c & a \end{bmatrix}

그렇다면 2 x 2가 아닌 n x n 행렬의 역행렬 공식은 어떻게 될까?

A1=1detACTA^{-1} = \frac{1}{\det A} C^T

위에서 CTC^T는 Cofactors matrix의 Transpose한 행렬이다.
위 2 x 2의 행렬의 예시를 보자.

C=[dcba],CT=[dbca]C=\begin{bmatrix} d & -c \\ -b & a \end{bmatrix}, C^T=\begin{bmatrix} d & -b \\ -c & a \end{bmatrix}

그렇다면 어떻게 위 역행렬 공식을 증명할 수 있을까?
이를 위해 ACT=(detA)IAC^T=(\det A)I 라는 걸 증명해보자.
(지난 강의에서 배운 detA=a11C11+a12C12+...+a1nC1n\det A = a_{11}C_{11}+a_{12}C_{12}+...+a_{1n}C_{1n}을 이용한다.)

[a11...a1n..........an1...ann][c11...c1nc12.........c1n...cnn]=[detA0....00detA..........detA00....0detA]\begin{bmatrix} a_{11} & ... & a_{1n} \\ . & ... & . \\ . & ... & . \\ a_{n1} & ... & a_{nn} \\ \end{bmatrix} \begin{bmatrix} c_{11} & ... & c_{1n} \\ c_{12} & ... & . \\ . & ... & . \\ c_{1n} & ... & c_{nn} \\ \end{bmatrix} = \begin{bmatrix} \det A & 0 & ... & . & 0 \\ 0 & \det A & ... & . & . \\ . & . & ... & \det A & 0 \\ 0 & . & ... & 0 & \det A \\ \end{bmatrix}

위와 같이 ACTAC^T의 결과가 (detA)I(\det A)I라는 것을 확인할 수가 있다.
왜 주대각 성분을 제외한 나머지가 0인지는 같은 행 혹은 열을 포함한 cofactor와 aija_{ij}의 연산은 최종적으로 다 합하면 결과가 0이 나오기 때문이다.
다음 예시를 통해 이해할 수 있다.

A=[abcd],CT=[dbca]A=\begin{bmatrix} a & b \\ c & d \end{bmatrix}, C^T=\begin{bmatrix} d & -b \\ -c & a \end{bmatrix}
ACT=[adbc00adbc]AC^T=\begin{bmatrix} ad-bc & 0 \\ 0 & ad-bc \end{bmatrix}

이제 2. Cramer's Rule for X=A1bX=A^{-1}b를 알아보자.
위에서 A1=1detACTA^{-1}=\frac{1}{\det A}C^T라는 공식을 알았다. 이를 적용한다.

X=A1b=1detACTbX=A^{-1}b=\frac{1}{\det A}C^Tb

그리고 Cramer's rule은 다음과 같다.

x1=detB1detAx_{1}=\frac{\det B_1}{\det A}
x2=detB2detAx_{2}=\frac{\det B_2}{\det A}
......
A with column 1 replaced by b.=B1=[bn1 columns of A]\text{$A$ with column $1$ replaced by $b$.}= B_1= \begin{bmatrix} | & | & | & | \\ b & & \text{$n-1$ columns of $A$} & \\ | & | & | & | \\ \end{bmatrix}

즉, B1B_1AA의 칼럼 중 1번 칼럼을 bb로 교체하고 나머지는 그대로 놔둔 행렬이다.

Bj=A with column j replaced by b.B_j=\text{$A$ with column $j$ replaced by $b$.}

(특별한 수학적 원리가 있기보다는 이러한 규칙을 발견했다는 데 의의가 있다.)

하지만 Cramer's Rule은 연산 비용이 매우 많이 들어 추천되지 않는 방법이다.


다음으로 3. detA=Volume of box|\det A|=\text{Volume of box} 를 알아보자.


detA=Volume of box|\det A|=\text{Volume of box}

위와 같이 3 x 3 행렬 AA의 determination을 그려보면 박스의 부피로 나오는 것을 알 수가 있다.

  1. 그렇다면 만약 A=IA=I라면 어떨까?
    • 이 경우, box는 unit cube 형태로 나올 것이다.
  2. 만약 A=QA=Q (QQ는 orthogonal matrixf를 의미)라면 어떨까?
    • 이 경우, box는 cube 형태로 나오며, 값은 1-1 또는 11일 것이다.
    • QTQ=IQ^TQ=I
    • det(QTQ)=detI=1\det (Q^TQ)=\det I = 1
    • detQTdetQ=1\det Q^T \det Q=1
    • (detQ)2=1(\det Q)^2=1
    • detQ=1 or1\therefore \det Q = 1 \ or -1

detA=Volume of box|\det A|=\text{Volume of box} 식이 지지난 강의에서 배운 properties of determinations에도 적용이 될까? 한 번 확인해보자.

  1. detI=1\det I = 1
    • 이건 (1,0,0),(0,1,0),(0,0,1)(1,0,0),(0,1,0),(0,0,1) 벡터를 그려보면,
      detA=1=1=Volume of box|\det A|=|1|=1=\text{Volume of box} 라는 걸 알 수 있다.
  2. 0110=1\begin{vmatrix}0 & 1 \\ 1 & 0\end{vmatrix}=-1
    • 마찬가지로 detA=1=1=Volume of box|\det A|=|-1|=1=\text{Volume of box} 라는 걸 알 수 있다.
  3. (a) tatbcd=tabcd\begin{vmatrix}ta & tb \\ c & d\end{vmatrix}=t\begin{vmatrix}a & b \\ c & d\end{vmatrix} (b) a+ab+bcd=abcd+abcd\begin{vmatrix}a+a' & b+b' \\ c & d\end{vmatrix}=\begin{vmatrix}a & b \\ c & d\end{vmatrix}+\begin{vmatrix}a' & b' \\ c & d\end{vmatrix}
    • (a) 벡터 (a,b)(a, b)tt만큼 곱해주면 벡터는 그만큼 더 나아간다. 그리고 역시 부피도 그만큼 증가할 것이다.
    • (a) 따라서 detAt=tdetA=t× Volume of box of A|\det A_t|=|t\det A|=\text{$t \times$ Volume of box of $A$} 라는 걸 알 수 있다.
    • (b)에 대해서는 시간상 확실한 설명은 없었다. 아마 두 개의 평면(박스)이 나올 것이고, 이 두 평면(박스)의 넓이(부피)를 더해주면 같다는 걸 증명할 수 있을 것 같다.
profile
천천히, 그리고 꾸준히.

0개의 댓글