티스토리 뷰

www.edwith.org/linearalgebra4ai/lecture/20102/

 

[LECTURE] 선형결합 : edwith

학습목표 이번 강의에서는 벡터들 간의 선형결합에 대한 개념과 벡터 공간 상의 span의 개념에 대해 알아보겠습니다. 그리고 선형결합과 관련하여 네 가지의 새로운 관점을 통해 행렬의... - 커��

www.edwith.org

Linear Combinations

Given vectors v1, v2, ,vp in Rn and given scalars c1, c2, ,cp,

c1v1+c2v2++cpxp

is called a linear combination of  v1, v2, ,vp with weights or coefficients c1, c2,,cp.

 

The weights in a linear combination can be any real numbers, including zero.

Span

Given a set of vectors v1, v2, ,vp in Rn, Span {v1, v2, ,vp} is defined as the set of all linear combinations of v1, v2, ,vp.

 

The collection of all vectors that can be written in the form

c1v1+c2v2++cpxp

with aribitrary scalars c1, c2, ,cp.

 

주어진 벡터들의 Linear Combination을 통해 만들 수 있는 모든 벡터들의 집합

From Matrix Equation to Vector Equation

60x1+5.5x2+1x3=66

65x1+5.0x2+0x3=74

55x1+6.0x2+1x3=78

 

위와 같은 Linear System이 있을 때, Matrix 연산으로 표현할 수 있습니다.

[605.51655.00556.01][x1x2x3]=[667478]

그리고 위 식에서 좌변을 Vector 연산으로 표현할 수 있습니다.

[606555]x1+[5.55.06.0]x2+[101]x3=[667478]

a1x1+a2x2+a3x3=b

where a1=[606555], a2=[5.55.06.0], a3=[101], and b=[667478].

 

좌변만 보았을때 Linear Combination 형태로 표현됨을 알 수 있습니다. Span의 개념을 잘 생각해보면 해당 Linear System을 만족하는 Solution(x1, x2, x3)의 존재 여부를 판단할 수 있습니다.

 

Span{a1, a2, a3}에 b가 존재할때, a1, a2, a3에 대한 Linear Combination으로 b를 만들어낼 수 있다는 의미이며, 이는 Solution x1, x2, x3가 존재한다는 의미이기 때문입니다. 반대의 경우 해당 Linear System은 Solution이 존재하지 않습니다.

Matrix Multiplications as Linear Combinations of Vectors

In spired by the vector equation, we can view \(A\mathbf{x}\) as a linear combination of columns of the left matrix:

[605.51655.00556.01][x1x2x3]=Ax=[a1a2a3][x1x2x3]=a1x1+a2x2+a3x3

Matrix Multiplications as Column Combinations

Linear combinations of columns

Left matrix: bases(벡터의 집합으로 생각), Right matrix: coefficients

One column on the right

[110101111][123]=[111]1+[101]2+[011]3

Left Matrix의 각 Column Vector를 Linear Combination을 구성하는 Vector로 볼 수 있으며, Right Vector의 각 원소를 Linear Combination의 Coefficients로 볼 수 있습니다. 그리고, 위 연산에 대해 Right Vector의 원소 값이 어떻게 변하든 결국 Left Matrix의 Column Vectors의 Span에 존재하게 됩니다.

Multi-columns on the right

[110101111][112031]=[x1y1x2y2x3y3]=[xy]

 

위 행렬곱을 Linear Combination 형태로 표현하면 아래의 식으로 표현할 수 있습니다.

x=[x1x2x3]=[111]1+[101]2+[011]3 

y=[y1y2y3]=[111](1)+[101]0+[011]1

 

Left Matrix의 각 Column Vector를 Linear Combination을 구성하는 Vector로 볼 수 있으며 Right Matrix의 각 원소를 Linear Combination의 Coefficients로 볼 수 있습니다. 그리고, x,y는 Right Matrix의 원소 값이 어떻게 변하든 결국 Left Matrix의 Column Vectors의 Span에 존재하게 됩니다.

Matrix Multiplications as Row Combinations

Linear combinations of rows of the right matrix

Right matrix: bases, Left matrix: coefficients

One row on the left

[123][110101111]=1×[110]+2×[101]+3×[111]

좌변의 Left Vector는 Coefficients, Right Matrix의 각 Row Vector는 Linear Combination을 이루는 Vector로 여길 수 있습니다.

Multi-columns on the left

[123101][110101111]=[x1x2x3y1y2y3]=[xTyT]

xT=[x1x2x3]=1[110]+2[101]+3[111]

yT=[y1y2y3]=1[110]+0[101]+(1)[111]

좌변의 Left Matrix를 Coefficients, Right Matrix의 각 Row Vector는 Linear Combination을 이루는 Vector로 여길 수 있습니다.

Matrix Multiplications as Sum of (Rank-1) Outer Products

(Rank-1) outer product

[111][123]=[123123123]

a=[111] 

b=[123]

Sum of (Rank-1) outer products

[111111][123456]=[111][123]+[111][456] 

a=[111],b=[123],c=[111],d=[456] 

[ac][bd]=ab+cd

행렬의 곱을 두 행렬의 Column Vectors와 Row Vectors간의 Outer Product와 그로부터 생성되는 Matrix의 합으로 표현할 수 있습니다.

 

PCA(Principal Component Analysis), SVD(Singular Vector Decomposition)등 Sum of (Rank-1) Outer Products의 개념을 기저에 두고 행렬을 특정 요소의 합으로 표현하는 방법이 많으므로 해당 내용을 이해하는 것이 중요합니다.

댓글
공지사항
최근에 올라온 글
최근에 달린 댓글
Total
Today
Yesterday
링크
«   2025/04   »
1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30
글 보관함