第 06 章

列空间与零空间:矩阵产生的两个子空间

列空间回答「哪些 b 有解」,零空间回答「哪些 x 被送到零」。前者由列向量张成,后者由齐次方程的解组成;两者都是子空间,而 Ax=b 的解集不是。

Column Space and Nullspace: The Two Subspaces a Matrix Creates

The column space answers which b are solvable; the nullspace answers which x are sent to zero. Both are subspaces; the solution set of Ax=b is not.

首发于 2009.05.07中 / EN
系统课程linear-algebracolumn-spacenullspacesubspace
语言
字号
目录
篇内目录

从子空间的两个判据说起

一个集合要成为子空间,只需满足两条:

  1. 对加法封闭:v,wv,w 在集合里,则 v+wv+w 也在;
  2. 对数乘封闭:vv 在集合里,则 cvcv 也在。

合起来就是:对线性组合封闭。另外,这两条自动保证零向量在集合里(取 c=0c=0)。

在 R3R^3 中,子空间只有四类:整个空间、过原点的平面、过原点的直线、只含零向量。

并集不是子空间,交集是

取一个过原点的平面 PP 和一条不在平面内的过原点直线 LL。

  • P∪LP\cup L 不是子空间:从平面上取一个向量,从直线上取一个向量,相加后通常既不在平面上也不在直线上。
  • P∩LP\cap L 是子空间:在这个例子里交集只有零向量,而 {0}\{\mathbf{0}\} 本身是子空间。

一般地,两个子空间的交集一定是子空间。设 S,TS,T 是子空间,v,w∈S∩Tv,w\in S\cap T:

  • v+w∈Sv+w\in S(因为 SS 是子空间),v+w∈Tv+w\in T(因为 TT 是子空间),所以 v+w∈S∩Tv+w\in S\cap T;
  • cv∈Scv\in S 且 cv∈Tcv\in T,所以 cv∈S∩Tcv\in S\cap T。

交集是「同时满足两组要求」的集合,要求越多,集合越小,但仍然保持封闭。

时间轴

时间内容
2:04三维中的平面与直线
4:50并集:把平面与直线放在一起
6:14并集是否仍是子空间
7:21交集是子空间
9:45交集的封闭性验证
12:46R4R^4 中的子空间
16:19Ax=bA\mathbf{x}=\mathbf{b} 是否总有解
26:22哪些列可以扔掉而列空间不变
29:12零空间的定义
31:10求 AA 的零空间
32:55零空间是向量空间
37:53两个解相加仍在零空间
43:53子空间必须过原点

列空间:由列向量张成

取

A=[112213314415].A= \begin{bmatrix} 1&1&2\\ 2&1&3\\ 3&1&4\\ 4&1&5 \end{bmatrix}.

这是 4×34\times3 矩阵,三列都是 R4R^4 中的向量:

c1=[1234],c2=[1111],c3=[2345].\mathbf{c}_1=\begin{bmatrix}1\\2\\3\\4\end{bmatrix},\quad \mathbf{c}_2=\begin{bmatrix}1\\1\\1\\1\end{bmatrix},\quad \mathbf{c}_3=\begin{bmatrix}2\\3\\4\\5\end{bmatrix}.

列空间 C(A)C(A) 是这三列的所有线性组合:

C(A)={x1c1+x2c2+x3c3∣x1,x2,x3∈R}.C(A)=\{x_1\mathbf{c}_1+x_2\mathbf{c}_2+x_3\mathbf{c}_3\mid x_1,x_2,x_3\in R\}.

它是 R4R^4 的子空间。为什么不是整个 R4R^4?因为三个向量最多张成三维;事实上这里连三维都不到,注意

c3=c1+c2,\mathbf{c}_3=\mathbf{c}_1+\mathbf{c}_2,

所以第三列没有带来新方向。列空间实际上是 R4R^4 中的一个二维子空间(一个过原点的平面)。

一句话总结: 列空间就是「所有可能的 AxA\mathbf{x}」构成的集合;它的维度由真正独立的列的个数决定。

与 Ax=bA\mathbf{x}=\mathbf{b} 的关系

这是本讲最重要的连接:

Ax=b 有解  ⟺  b∈C(A).A\mathbf{x}=\mathbf{b}\ \text{有解} \iff \mathbf{b}\in C(A).

理由很直接:AxA\mathbf{x} 按定义就是列的组合,系数是 x\mathbf{x} 的分量。所以:

  • 若 b\mathbf{b} 是列的组合,那组系数就是一个解;
  • 若 b\mathbf{b} 不是列的组合,就不存在解。

对这个 4×34\times3 矩阵,四个方程只有三个未知数,所以大多数 b\mathbf{b} 无解。但有解的 b\mathbf{b} 恰好构成列空间这个平面。

具体例子:

  • b=[1,2,3,4]T=c1\mathbf{b}=[1,2,3,4]^T=\mathbf{c}_1,解是 (1,0,0)(1,0,0);
  • b=[1,1,1,1]T=c2\mathbf{b}=[1,1,1,1]^T=\mathbf{c}_2,解是 (0,1,0)(0,1,0);
  • 任意 ac1+bc2a\mathbf{c}_1+b\mathbf{c}_2,解是 (a,b,0)(a,b,0)。

零空间:被送到零的 xx

同一个矩阵,换一个完全不同的子空间。零空间 N(A)N(A) 是所有满足

Ax=0A\mathbf{x}=\mathbf{0}

的 x\mathbf{x} 组成的集合。

注意它装的是 xx,不是 bb。对这个 4×34\times3 矩阵,x\mathbf{x} 有三个分量,所以 N(A)N(A) 是 R3R^3 的子空间。

求出这个例子的零空间

观察第三列等于前两列之和:

c1+c2−c3=0.\mathbf{c}_1+\mathbf{c}_2-\mathbf{c}_3=\mathbf{0}.

所以

x=[11−1]\mathbf{x}=\begin{bmatrix}1\\1\\-1\end{bmatrix}

是一个解。任何倍数也是解:

N(A)={c[11−1] | c∈R}.N(A)=\left\{c\begin{bmatrix}1\\1\\-1\end{bmatrix}\ \middle|\ c\in R\right\}.

几何上,这是 R3R^3 中过原点的一条直线。

为什么零空间一定是子空间

设 Av=0A\mathbf{v}=\mathbf{0},Aw=0A\mathbf{w}=\mathbf{0}。则

A(v+w)=Av+Aw=0+0=0,A(\mathbf{v}+\mathbf{w})=A\mathbf{v}+A\mathbf{w}=\mathbf{0}+\mathbf{0}=\mathbf{0},

所以 v+w\mathbf{v}+\mathbf{w} 也在零空间。对数乘同理:

A(cv)=cAv=c0=0.A(c\mathbf{v})=cA\mathbf{v}=c\mathbf{0}=\mathbf{0}.

这里用到矩阵乘法的分配律与数乘可交换。零空间因此对线性组合封闭,是子空间。

关键对比:Ax=bA\mathbf{x}=\mathbf{b} 的解集不是子空间

把右端换成 b=[1,2,3,4]T\mathbf{b}=[1,2,3,4]^T。它有一个解 (1,0,0)(1,0,0),还有另一个解 (0,−1,1)(0,-1,1):

0⋅c1+(−1)c2+1⋅c3=−c2+c3=[1234].0\cdot\mathbf{c}_1+(-1)\mathbf{c}_2+1\cdot\mathbf{c}_3 =-\mathbf{c}_2+\mathbf{c}_3 =\begin{bmatrix}1\\2\\3\\4\end{bmatrix}.

但这个解集不是子空间,因为零向量不是解:A0=0≠bA\mathbf{0}=\mathbf{0}\neq\mathbf{b}。

它的结构是「一个特解 + 零空间」:

x=[100]+c[11−1].\mathbf{x}= \begin{bmatrix}1\\0\\0\end{bmatrix} +c\begin{bmatrix}1\\1\\-1\end{bmatrix}.

几何上,这是 R3R^3 中一条不过原点的直线。所有解构成零空间的一条平移,称为仿射集,而不是子空间。

补充: 这也解释了为什么齐次方程 Ax=0A\mathbf{x}=\mathbf{0} 特别重要:只有它的解集包含原点,因而才可能是子空间;非齐次方程的解集是它的一次平移。

描述子空间的两种方式

这一讲给出了构造子空间的两种基本语言:

方式例子特征
给出一组向量,取所有线性组合列空间「由内向外」张成
给出一组方程,取所有解零空间「由外向内」约束
  • 列空间:告诉我几列,我取组合,得到一个子空间。
  • 零空间:告诉我方程,我取解,得到一个子空间。

两种方式都会在后续课程中反复出现:列空间对应「像」,零空间对应「核」。

本讲结论

  • 子空间对线性组合封闭,必含零向量。
  • 两个子空间的交集仍是子空间;并集一般不是。
  • 列空间 C(A)C(A) 是列的所有组合,位于 RmR^m(mm 为行数)。
  • Ax=bA\mathbf{x}=\mathbf{b} 有解当且仅当 b∈C(A)\mathbf{b}\in C(A)。
  • 零空间 N(A)N(A) 是 Ax=0A\mathbf{x}=\mathbf{0} 的所有解,位于 RnR^n(nn 为列数)。
  • 零空间是子空间;非齐次方程的解集不是,它是零空间的一次平移。
  • 列数多于独立列数时,列空间维度下降;本例中第三列是前两列之和,列空间是二维的。

理解检验

1. 证明两个子空间的交集是子空间,并给出并集不是子空间的反例。

参考答案:设 S,TS,T 是子空间,v,w∈S∩T\mathbf{v},\mathbf{w}\in S\cap T。由 SS 是子空间,v+w∈S\mathbf{v}+\mathbf{w}\in S;由 TT 是子空间,v+w∈T\mathbf{v}+\mathbf{w}\in T,所以 v+w∈S∩T\mathbf{v}+\mathbf{w}\in S\cap T。数乘同理:cv∈Sc\mathbf{v}\in S 且 cv∈Tc\mathbf{v}\in T。封闭性成立,交集是子空间(空集情形另论,通常指非空交集)。 并集的反例:在 R3R^3 中取平面 P={(x,y,0)}P=\{(x,y,0)\} 与直线 L={(0,0,z)}L=\{(0,0,z)\}。u=(1,0,0)∈P\mathbf{u}=(1,0,0)\in P,w=(0,0,1)∈L\mathbf{w}=(0,0,1)\in L,但

u+w=(1,0,1),\mathbf{u}+\mathbf{w}=(1,0,1),

它的第三个分量不为 00,所以不在 PP;第一个分量不为 00,所以不在 LL。加法把向量带出了并集,因此 P∪LP\cup L 不是子空间。

2. 为什么本例的列空间不是整个 R4R^4?用「三个向量最多张成几维」回答。

参考答案:AA 是 4×34\times3,只有三个列向量,三个向量最多张成三维,永远到不了四维的 R4R^4。本例更弱:第三列等于前两列之和,c3=c1+c2\mathbf{c}_3=\mathbf{c}_1+\mathbf{c}_2,所以真正独立的方向只有两个,列空间是一个二维子空间(R4R^4 中过原点的平面)。因此绝大多数 b∈R4\mathbf{b}\in R^4 落在平面外,对应的方程无解。

3. 写出本例的零空间,并验证 c1+c2−c3=0\mathbf{c}_1+\mathbf{c}_2-\mathbf{c}_3=\mathbf{0}。

参考答案:验证列的组合:

[1234]+[1111]−[2345]=[1+1−22+1−33+1−44+1−5]=[0000].\begin{bmatrix}1\\2\\3\\4\end{bmatrix}+\begin{bmatrix}1\\1\\1\\1\end{bmatrix}-\begin{bmatrix}2\\3\\4\\5\end{bmatrix} =\begin{bmatrix}1+1-2\\2+1-3\\3+1-4\\4+1-5\end{bmatrix} =\begin{bmatrix}0\\0\\0\\0\end{bmatrix}.

所以 x=(1,1,−1)T\mathbf{x}=(1,1,-1)^T 满足 Ax=0A\mathbf{x}=\mathbf{0},任何倍数也满足。零空间是

N(A)={c[11−1] | c∈R},N(A)=\left\{c\begin{bmatrix}1\\1\\-1\end{bmatrix}\ \middle|\ c\in R\right\},

即 R3R^3 中过原点的一条直线。

4. 为什么 Ax=bA\mathbf{x}=\mathbf{b}(b≠0\mathbf{b}\ne\mathbf{0})的解集不是子空间?

参考答案:子空间必须含零向量,而 A0=0≠bA\mathbf{0}=\mathbf{0}\ne\mathbf{b},所以零向量不是解,解集不含原点,直接出局。也可以从封闭性看:若 x1,x2\mathbf{x}_1,\mathbf{x}_2 都是解,则

A(x1+x2)=b+b=2b≠bA(\mathbf{x}_1+\mathbf{x}_2)=\mathbf{b}+\mathbf{b}=2\mathbf{b}\ne\mathbf{b}

(因为 b≠0\mathbf{b}\ne\mathbf{0}),两个解相加不再是解。所以解集对加法也不封闭。它的真实结构是「特解 + 零空间」,即零空间的一次平移。

5. 已知一个特解与零空间,如何写出全部解?

参考答案:设 xp\mathbf{x}_p 是一个特解,N(A)N(A) 是零空间,则全部解为

x=xp+xn,xn∈N(A).\mathbf{x}=\mathbf{x}_p+\mathbf{x}_n,\qquad \mathbf{x}_n\in N(A).

两个方向都要验证:一方面,A(xp+xn)=b+0=bA(\mathbf{x}_p+\mathbf{x}_n)=\mathbf{b}+\mathbf{0}=\mathbf{b},所以每个这样的向量都是解;另一方面,若 x\mathbf{x} 也是解,则 A(x−xp)=b−b=0A(\mathbf{x}-\mathbf{x}_p)=\mathbf{b}-\mathbf{b}=\mathbf{0},即 x−xp∈N(A)\mathbf{x}-\mathbf{x}_p\in N(A),所以它一定写成这种形式,没有遗漏。在本例中就是

x=[100]+c[11−1].\mathbf{x}=\begin{bmatrix}1\\0\\0\end{bmatrix}+c\begin{bmatrix}1\\1\\-1\end{bmatrix}.

与前后课的关系

这一讲定义了矩阵的两个核心子空间。下一讲用消元系统地求出零空间:主元列与自由列的分工,会给出零空间的一组「特殊解」,并由此得到维数公式。

Starting from the two subspace tests

A set is a subspace when it satisfies two conditions:

  1. closed under addition: if v,wv,w are in the set, so is v+wv+w;
  2. closed under scalar multiplication: if vv is in the set, so is cvcv.

Together: closed under linear combinations. These two conditions also force the zero vector into the set (take c=0c=0).

In R3R^3 the subspaces are exactly: the whole space, a plane through the origin, a line through the origin, and the zero vector alone.

Union fails, intersection works

Take a plane PP through the origin and a line LL through the origin that is not inside PP.

  • P∪LP\cup L is not a subspace: take a vector from the plane and one from the line; their sum is usually in neither.
  • P∩LP\cap L is a subspace: here the intersection is only the zero vector, and {0}\{\mathbf{0}\} is a subspace.

In general, the intersection of two subspaces is a subspace. Let S,TS,T be subspaces and v,w∈S∩Tv,w\in S\cap T:

  • v+w∈Sv+w\in S because SS is a subspace, and v+w∈Tv+w\in T because TT is a subspace, so v+w∈S∩Tv+w\in S\cap T;
  • cv∈Scv\in S and cv∈Tcv\in T, so cv∈S∩Tcv\in S\cap T.

An intersection imposes both sets of requirements, so it is smaller, but it stays closed.

Timeline

TimeTopic
2:04Planes and lines in three dimensions
4:50The union of a plane and a line
6:14Is the union still a subspace?
7:21The intersection is a subspace
9:45Checking closure of the intersection
12:46Subspaces inside R4R^4
16:19Does Ax=bA\mathbf{x}=\mathbf{b} have a solution for every b\mathbf{b}?
26:22Which columns can be thrown away?
29:12The definition of the null space
31:10Computing the null space
32:55The null space is a vector space
37:53The sum of two solutions stays in the null space
43:53Subspaces must pass through the origin

The column space: spanned by the columns

Take

A=[112213314415].A= \begin{bmatrix} 1&1&2\\ 2&1&3\\ 3&1&4\\ 4&1&5 \end{bmatrix}.

This is 4×34\times3; its three columns live in R4R^4:

c1=[1234],c2=[1111],c3=[2345].\mathbf{c}_1=\begin{bmatrix}1\\2\\3\\4\end{bmatrix},\quad \mathbf{c}_2=\begin{bmatrix}1\\1\\1\\1\end{bmatrix},\quad \mathbf{c}_3=\begin{bmatrix}2\\3\\4\\5\end{bmatrix}.

The column space C(A)C(A) is all linear combinations:

C(A)={x1c1+x2c2+x3c3∣x1,x2,x3∈R}.C(A)=\{x_1\mathbf{c}_1+x_2\mathbf{c}_2+x_3\mathbf{c}_3\mid x_1,x_2,x_3\in R\}.

It is a subspace of R4R^4. Why not all of R4R^4? Three vectors span at most three dimensions. In fact here they span fewer, because

c3=c1+c2,\mathbf{c}_3=\mathbf{c}_1+\mathbf{c}_2,

so the third column adds no new direction. The column space is a two-dimensional subspace of R4R^4 — a plane through the origin.

One-sentence summary: The column space is the set of all possible AxA\mathbf{x}; its dimension is the number of genuinely independent columns.

The connection to Ax=bA\mathbf{x}=\mathbf{b}

This is the lecture's central link:

Ax=b is solvable  ⟺  b∈C(A).A\mathbf{x}=\mathbf{b}\ \text{is solvable} \iff \mathbf{b}\in C(A).

The reason is immediate: AxA\mathbf{x} is by definition a combination of the columns, with the entries of x\mathbf{x} as coefficients. So:

  • if b\mathbf{b} is a combination of the columns, those coefficients give a solution;
  • if not, no solution exists.

For this 4×34\times3 matrix there are four equations and only three unknowns, so most b\mathbf{b} are unsolvable. The solvable ones form exactly the plane that is the column space.

Examples:

  • b=[1,2,3,4]T=c1\mathbf{b}=[1,2,3,4]^T=\mathbf{c}_1 has solution (1,0,0)(1,0,0);
  • b=[1,1,1,1]T=c2\mathbf{b}=[1,1,1,1]^T=\mathbf{c}_2 has solution (0,1,0)(0,1,0);
  • any ac1+bc2a\mathbf{c}_1+b\mathbf{c}_2 has solution (a,b,0)(a,b,0).

The nullspace: the xx sent to zero

For the same matrix, a completely different subspace. The nullspace N(A)N(A) is the set of all x\mathbf{x} satisfying

Ax=0.A\mathbf{x}=\mathbf{0}.

It contains xx, not bb. Since x\mathbf{x} has three components, N(A)N(A) is a subspace of R3R^3.

Computing it for this example

Observe that the third column is the sum of the first two:

c1+c2−c3=0.\mathbf{c}_1+\mathbf{c}_2-\mathbf{c}_3=\mathbf{0}.

Therefore

x=[11−1]\mathbf{x}=\begin{bmatrix}1\\1\\-1\end{bmatrix}

is a solution, and so is every multiple:

N(A)={c[11−1] | c∈R}.N(A)=\left\{c\begin{bmatrix}1\\1\\-1\end{bmatrix}\ \middle|\ c\in R\right\}.

Geometrically, this is a line through the origin in R3R^3.

Why the nullspace must be a subspace

Let Av=0A\mathbf{v}=\mathbf{0} and Aw=0A\mathbf{w}=\mathbf{0}. Then

A(v+w)=Av+Aw=0+0=0,A(\mathbf{v}+\mathbf{w})=A\mathbf{v}+A\mathbf{w}=\mathbf{0}+\mathbf{0}=\mathbf{0},

so v+w\mathbf{v}+\mathbf{w} is also in the nullspace. For scalars,

A(cv)=cAv=c0=0.A(c\mathbf{v})=cA\mathbf{v}=c\mathbf{0}=\mathbf{0}.

The nullspace is closed under linear combinations, hence a subspace.

The key contrast: solutions of Ax=bA\mathbf{x}=\mathbf{b} are not a subspace

Change the right-hand side to b=[1,2,3,4]T\mathbf{b}=[1,2,3,4]^T. One solution is (1,0,0)(1,0,0); another is (0,−1,1)(0,-1,1):

0⋅c1+(−1)c2+1⋅c3=−c2+c3=[1234].0\cdot\mathbf{c}_1+(-1)\mathbf{c}_2+1\cdot\mathbf{c}_3 =-\mathbf{c}_2+\mathbf{c}_3 =\begin{bmatrix}1\\2\\3\\4\end{bmatrix}.

But the solution set is not a subspace, because the zero vector is not a solution: A0=0≠bA\mathbf{0}=\mathbf{0}\neq\mathbf{b}.

Its structure is "one particular solution plus the nullspace":

x=[100]+c[11−1].\mathbf{x}= \begin{bmatrix}1\\0\\0\end{bmatrix} +c\begin{bmatrix}1\\1\\-1\end{bmatrix}.

Geometrically, this is a line in R3R^3 that does not pass through the origin. The solution set is a translate of the nullspace — an affine set, not a subspace.

Background: This is why the homogeneous equation Ax=0A\mathbf{x}=\mathbf{0} matters so much. Only its solution set contains the origin, so only it can be a subspace; the nonhomogeneous solution set is a translate of it.

Two ways to describe a subspace

The lecture names the two basic constructions:

WayExampleCharacter
Give vectors, take all combinationsColumn spacespanning from inside
Give equations, take all solutionsNullspaceconstraining from outside
  • Column space: here are some columns; take their combinations.
  • Nullspace: here are the equations; take the solutions.

Both recur throughout the course: the column space is the image, and the nullspace is the kernel.

Takeaways

  • A subspace is closed under linear combinations and therefore contains zero.
  • The intersection of two subspaces is a subspace; the union generally is not.
  • The column space C(A)C(A) is all combinations of the columns, living in RmR^m.
  • Ax=bA\mathbf{x}=\mathbf{b} is solvable exactly when b∈C(A)\mathbf{b}\in C(A).
  • The nullspace N(A)N(A) is all solutions of Ax=0A\mathbf{x}=\mathbf{0}, living in RnR^n.
  • The nullspace is a subspace; the solution set of a nonhomogeneous system is a translate of it.
  • More columns than independent columns lowers the dimension of the column space; here the third column is the sum of the first two, so the column space is two-dimensional.

Check your understanding

1. Prove that the intersection of two subspaces is a subspace, and give an example showing the union need not be.

Reference answer: Let S,TS,T be subspaces and v,w∈S∩T\mathbf{v},\mathbf{w}\in S\cap T. Since SS is a subspace, v+w∈S\mathbf{v}+\mathbf{w}\in S; since TT is a subspace, v+w∈T\mathbf{v}+\mathbf{w}\in T; hence v+w∈S∩T\mathbf{v}+\mathbf{w}\in S\cap T. The same holds for scaling: cv∈Sc\mathbf{v}\in S and cv∈Tc\mathbf{v}\in T. So the intersection is closed and is a subspace. For the union, take in R3R^3 the plane P={(x,y,0)}P=\{(x,y,0)\} and the line L={(0,0,z)}L=\{(0,0,z)\}. Then u=(1,0,0)∈P\mathbf{u}=(1,0,0)\in P and w=(0,0,1)∈L\mathbf{w}=(0,0,1)\in L, but

u+w=(1,0,1),\mathbf{u}+\mathbf{w}=(1,0,1),

which is in neither PP (third coordinate nonzero) nor LL (first coordinate nonzero). The sum escapes the union, so P∪LP\cup L is not a subspace.

2. Why is this column space not all of R4R^4? Answer in terms of how many dimensions three vectors can span.

Reference answer: AA is 4×34\times3, so it has only three columns, and three vectors span at most three dimensions — never all of four-dimensional R4R^4. Here it is even smaller: c3=c1+c2\mathbf{c}_3=\mathbf{c}_1+\mathbf{c}_2, so only two independent directions remain and the column space is a two-dimensional plane through the origin. Most b∈R4\mathbf{b}\in R^4 lie off that plane, and those systems have no solution.

3. Write the nullspace of this matrix and verify c1+c2−c3=0\mathbf{c}_1+\mathbf{c}_2-\mathbf{c}_3=\mathbf{0}.

Reference answer: Compute the combination:

[1234]+[1111]−[2345]=[0000].\begin{bmatrix}1\\2\\3\\4\end{bmatrix}+\begin{bmatrix}1\\1\\1\\1\end{bmatrix}-\begin{bmatrix}2\\3\\4\\5\end{bmatrix} =\begin{bmatrix}0\\0\\0\\0\end{bmatrix}.

So x=(1,1,−1)T\mathbf{x}=(1,1,-1)^T solves Ax=0A\mathbf{x}=\mathbf{0}, and so does every multiple. The nullspace is

N(A)={c[11−1] | c∈R},N(A)=\left\{c\begin{bmatrix}1\\1\\-1\end{bmatrix}\ \middle|\ c\in R\right\},

a line through the origin in R3R^3.

4. Why is the solution set of Ax=bA\mathbf{x}=\mathbf{b} with b≠0\mathbf{b}\neq\mathbf{0} not a subspace?

Reference answer: A subspace must contain the zero vector, but A0=0≠bA\mathbf{0}=\mathbf{0}\ne\mathbf{b}, so zero is not a solution and the set misses the origin. Closure also fails: if x1,x2\mathbf{x}_1,\mathbf{x}_2 are solutions,

A(x1+x2)=b+b=2b≠bA(\mathbf{x}_1+\mathbf{x}_2)=\mathbf{b}+\mathbf{b}=2\mathbf{b}\ne\mathbf{b}

for b≠0\mathbf{b}\ne\mathbf{0}, so the sum is not a solution. The set is a translate of the nullspace, not a subspace.

5. Given one particular solution and the nullspace, how do you write all solutions?

Reference answer: With xp\mathbf{x}_p a particular solution and N(A)N(A) the nullspace,

x=xp+xn,xn∈N(A).\mathbf{x}=\mathbf{x}_p+\mathbf{x}_n,\qquad \mathbf{x}_n\in N(A).

Both directions hold: A(xp+xn)=b+0=bA(\mathbf{x}_p+\mathbf{x}_n)=\mathbf{b}+\mathbf{0}=\mathbf{b}, so each such vector is a solution; and if x\mathbf{x} is any solution, A(x−xp)=0A(\mathbf{x}-\mathbf{x}_p)=\mathbf{0}, so x−xp∈N(A)\mathbf{x}-\mathbf{x}_p\in N(A). Nothing is missing. In this example,

x=[100]+c[11−1].\mathbf{x}=\begin{bmatrix}1\\0\\0\end{bmatrix}+c\begin{bmatrix}1\\1\\-1\end{bmatrix}.

How it connects

This lecture defines the two central subspaces of a matrix. The next uses elimination to compute the nullspace systematically: pivot columns and free columns produce the "special solutions", and from them comes the dimension formula.

来源