9 분 소요

이번 포스트에서는 tranformation에서의 kernel과 range에 대해서 알아보겠습니다.


1) Kernel of transformationPermalink


(1) KernelPermalink


Definition : Kernel of transformation

If T:RnRmT:RnRm is a transformation, then the set of vectors in RnRn that TT maps into 00 is called kernel of TT and is denoted by ker(T)ker(T)

Ker(T)={xT(x)=0}Ker(T)={xT(x)=0}

즉 transformation T의 kernel은 T(x)=0을 만족시키는 모든 x 을 모은 집합입니다.


example

A=[112534]

에 대해서 matrix transformation TA의 kernel은

TA(x)=Ax=0

을 만족시키는 x의 집합니다. 따라서

[110250340]

다음의 augmented matrix를 가진 linear system을 푸는 문제로 바뀌고, 이를 풀게 되면

[100010000]

이 되어, 위 system의 solution은 zero vector밖에 존재하지 않아

ker(TA)={0}

가 됩니다.


example

정의역과 공역이 Rn인 Zero operator T0의 kernel은

ker(T0)={xT0(x)=0x=0}

을 만족시키는 x 집합이므로, Rn에 속하는 모든 벡터가 kernel에 속합니다. 따라서

ker(T0)=Rn

이 됩니다.


example

정의역과 공역이 Rn인 identity operator TI의 kernel은

ker(TI)={xTI(x)=Ix=0}

이므로, zero vector만 성립됩니다. 따라서

ker(TI)={0}

가 됩니다.

Kernel과 linear transformation과 관련된 정리는 다음과 같습니다.


Theorem

The kernel of a linear transformation always contains the zero vector

linear transformation의 kernel은 반드시 zero vector를 포함합니다.


Theorem

If T:RnRm is a linear transformation, then the kernel of T is a subspace of Rn

linear transformation의 kernel은 domain의 subspace가 됩니다.


(2) One to onePermalink


Definition : One to one

The transformation T:RnRm is one to one if T maps distinct vectors in Rn into distinct vectors in Rm

one to one은 함수에서 정의되는 일대일 함수의 정의와 같습니다. 즉

if  T(x)= T(y)x=y  for  all  x,yRn

일 때, T는 one to one이라고 합니다. 위 정의의 대우인

if  xyT(x) T(y)  for  all  x,yRn

또한 많이 사용됩니다.

one to one과 linear transformation, kernel은 다음과 같은 관계를 가지고 있습니다.


Theorem

If T:RnRm is a linear transformation, then the followings are equivalent

  1. T is one to one
  2. ker(T)=0

만약 T가 linear transformation이면, T가 one to one임과 kernel이 zero vector만 있는 것은 동치입니다.


Theorem

If A is m×n matrix, then the corresponding linear transformation TA:RnRm is one to one if and only if the linear system Ax=0 has only the trivial solution.

matrix transformation과 one to one 간의 관계를 나타내는 정리입니다. 위의 linear transformation이 one to one이면 kernel이 zero vector만을 가지는 것을 이용하면 쉽게 증명할 수 있습니다.


2) Range of transformationPermalink


(1) Range of transformationPermalink


Definition : Range of transformation

If T:RnRm is a transformation, then the range(image) of T, denoted by ran(T) is the set of all vectors in Rm that are images of at least one vector in Rn

range는 함수에서 정의되는 치역과 같습니다. 즉,

ran(T)={bb=T(x)  for  all  xRn}

kernel과 달리 range는 transfrom된 output 값들의 집합이기 때문에, 공역의 부분집합입니다.


example

A=[112534]

일 때, matrix transformation TA의 range는

ran(TA)={TA(x)xR2}

입니다. 이 때,

TA(x)=x1[123]+x2[154]

가 되고, R2에 포함되는 모든 x에 대해 위 output을 전부 모은 집합이니, 이는 A의 column의 linear combination을 모두 모은 집합입니다. 즉

ran(TA)=span{[123],[154]}


example

정의역과 공역이 Rn인 zero operator의 경우

ran(T0)={T0(x)xRn}={0}

output이 zero vector밖에 없기 때문에, range는 zero vector만을 가집니다.


example

정의역과 공역이 Rn인 identity operator TI의 range는

ran(TI)={TI(x)xRn}=Rn

Rn 에 속한 모든 vector x에 대해 output은 자기 자신이고, 이를 모두 모은 집합이니, range는 Rn이 됩니다.

Range와 linear transformation은 다음과 같은 관계를 가집니다.


Theorem

If T:RnRm is a linear transformation, then ran(T) is a subspace of Rm

linear transformation의 range는 공역의 subspace가 됩니다.


(2) OntoPermalink


**Definition : Onto **

A transformation T:RnRm is onto if the range of T is the entire codomain Rm

즉, range와 codomain이 같을 때 transformation은 onto라고 합니다.

onto와 range, linear transformation과의 관계는 다음과 같습니다.


Theorem

If A is ans m×n matrix, then the corresponding linear treansformation TA:RnRm is onto if and only if the linear system Ax=b is consistent form every b in Rm

linear transformation일 때 range의 성질과 onto의 정의를 이용하면 쉽게 증명할 수 있습니다.


3) Linear operatorPermalink


(1) Linear operatorPermalink

앞서서 정의한 linear transformation, kernel, one to one, range, onto를 linear operator에 적용하면 다음의 정리를 얻습니다.


**Theorem **

If T:RnRn is a linear operator on Rn, then T is one to one if and only if it is onto.


(2) Invertible Matrix TheoremPermalink

linear operator에서 onto, one to one의 성질을 이용하면, linear operator의 standard matrix의 invertibility에 대해서도 논할 수 있습니다. 만약 standard matrix가 one to one이고 onto이면(linear operator에서는 동치입니다. ), 해당 standard matrix는 invertible 합니다. 해당 명제의 역 또한 성립하구요. 따라서 이전의 Invertible Matrix Theorem에 다음의 명제가 추가됩니다.

Let A be a square n×n matrix. Then the following statements are equivalent. That is, for given A, the statements are either all true or all false

a. A is an invertible matrix

b. A is row equivalent to the n×n identity matrix

c. A has n pivot positions

d. The equation Ax=0 has only the trivial solution

e. The columns of A form a linearly independent set

f. The columns of A span Rn

g. There is an n×n matrix C such taht CA=I

h. There is an n×n matrix D such that AD=I

i. The equation Ax=0 has at least one solution for each b in Rn

j. AT is an invertible matrix

k. detA0

l. TA is one to one

m. TA is onto

지금까지 kernel과 range에 대해서 알아보았습니다. 다음 포스트에서는 matrix로 정의할 수 있는 vector space인 row space, column space, null space에 대해서 알아보겠습니다. 질문이나 오류 있으면 댓글 남겨주세요! 감사합니다!


Appendix : Proof of TheoremPermalink


(1) Kernel , one to onePermalink


Theorem

The kernel of a linear transformation always contains the zero vector

  • proof

T:RnRm 가 linear transformation일 때, T의 kernel은

Ker(T)={xT(x)=0}

입니다. T가 linear transformation이므로

T(0)=T(0 v)=0T(v)=0

을 만족합니다. 따라서 zero vector는 linear transformation의 kernel에 반드시 포함됩니다.


Theorem

If T:RnRm is a linear transformation, then the kernel of T is a subspace of Rn

  • proof

Kernel이 Rn의 subspace임을 확인하기 위해서는 4가지를 확인해야 합니다.

  1. Ker(T)Rn

    Kernel의 정의에 따라 Rn의 subspace입니다.

  2. 0Ker(T)

    앞선 정리에서 linear transformation의 kernel에는 zero vector를 반드시 포함합니다.

  3. u,vKer(T)T(u)=T(v)=0

    이므로, T(u+v)=T(u)+T(v)=0을 만족하여 u+vKer(T)을 만족합니다.

  4. uKer(T),cR

    일 때, T(cu)=cT(u)=0이 되어 cuKer(T)를 만족합니다.

따라서 subspace의 조건을 모두 만족하므로 Ker(T)Rn의 subspace입니다.


Theorem

If T:RnRm is a linear transformation, then the followings are equivalent

  1. T is one to one
  2. ker(T)=0
  • proof

T is one to one Ker(T)=0

T가 one to one이면

T(x)=T(y)x=y

을 만족합니다. 또한 T가 linear transformation이므로 0T의 kernel에 속합니다. 만약 0 가 아닌 다른 벡터 vKer(T)에 속한다고 가정해봅시다.

그럼

T(v)=T(0)=0

이고, T가 one to one이므로

v=0

가 됩니다. 현재 v0가 아니라고 가정하였기 때문에 모순이 발생하여, ker(T)=0입니다.

Ker(T)=0 T is one to one

T가 one to one임을 확인하기 위하여

T(x)=T(y)

인 경우를 생각해봅시다. 이는 T가 linear transformation이므로

T(x)T(y)=T(xy)=0

가 성립합니다. 즉 xyKer(T)이고, Ker(T)에 속한 vector는 0이므로

xy=0x=y

따라서 T는 one to one이 됩니다.


Theorem

If A is m×n matrix, then the corresponding linear transformation TA:RnRm is one to one if and only if the linear system Ax=0 has only the trivial solution.

  • proof

TA가 one to one이면 TA의 kernel은 zero vector만 존재합니다. 즉

Ker(TA)={xTA(x)=Ax=0}={0}

을 만족시키는 x0밖에 없기 때문에, linear system Ax=0 은 trivial solution만을 가지게 됩니다.

반대로 linear system Ax=0 이 trivial solution만을 가지면 TA의 kernel이 zero vector만 가지기 때문에, one to one이 성립됩니다.


(2) Range, ontoPermalink


Theorem

If T:RnRm is a linear transformation, then ran(T) is a subspace of Rm

  • proof

T의 range가 Rm의 subspace임을 밝히기 위해서는 4가지를 확인해야 합니다.

  1. ran(T)Rm

    range의 정의에 의해 성립합니다.

  2. 0ran(T)

    T가 linear transformation이므로 T(0)=0임을 만족하고, 따라서 0ran(T)를 만족합니다.

  3. u,vran(T) 이면, 어떤 x,yRn이 존재하여 T(x)=u,T(y)=v을 만족합니다. 이 때

    u+v=T(x)+T(y)=T(x+y)ran(T)입니다.

  4. uran(T),cR에 대해서, T(x)=u을 만족하는 x가 존재합니다. 따라서

    cu=cT(x)=T(cx)ran(T) 입니다.

따라서 ran(T)Rm의 subspace입니다.


Theorem

If A is ans m×n matrix, then the corresponding linear treansformation TA:RnRm is onto if and only if the linear system Ax=b is consistent for every b in Rm

  • proof

TA:RnRm 가 onto이므로, ran(TA)=Rm입니다. range의 정의가

ran(TA)={TA(x)xRn}=Rm

이므로, Rm에 존재하는 임의의 vector b에 대해 TA(x)=Ax를 만족하는 x존재합니다. 즉

Ax=b

는 모든 b에 대해 solution을 가집니다. 즉 consistent합니다.

마찬가지로, Ax=b가 모든 bRm에 대해 consistent하면 range 정의에 따라 ran(TA)=Rm입니다. 따라서 TA는 onto입니다.


(3) Linear operatorPermalink


**Theorem **

If T:RnRn is a linear operator on Rn, then T is one to one if and only if it is onto.

  • proof

T가 one to one이면 T의 standard matrix [T]에 대해서

[T]x=0

이 반드시 trivial solution을 가집니다. 이는 invertible matrix theorem에 따라 [T]는 invertible합니다.

[T]가 invertible하면, Rn에 속하는 모든 b에 대해

[T]x=b

가 consistent합니다. 따라서

ran(T)=Rn

이므로 T는 onto입니다.

위 과정을 거꾸로 진행하면 T가 onto이면 T가 one to one인 것 또한 쉽게 알 수 있습니다.

댓글남기기