第 13 章

第一次测验复习:按题型整理

把前半学期的考点压成几类:维数与秩的计数、由完整解反推矩阵、真伪判断、乘积的零空间,以及行空间与零空间的交集。

Quiz 1 Review: Organized by Problem Type

The first half of the course compressed into a few problem types: dimension and rank counts, reconstructing a matrix from its complete solution, true or false, null spaces of products, and the intersection of row and null spaces.

首发于 2009.05.07中 / EN
系统课程linear-algebraexam-reviewranknull-space
语言
字号
目录
篇内目录

这次复习覆盖什么

第一次测验考的是 Ax=bA\mathbf{x}=\mathbf{b} 这一半内容,重点是矩形矩阵:零空间、左零空间、秩、四个基本子空间。矩阵可逆时这些概念都「自动成立」,一旦变成矩形,就必须逐个判断。

下面按题型整理,而不是按课堂顺序复述。

时间轴

时间内容
0:09复习开场:本次覆盖的范围
1:49题型一:张成空间的维数
5:29行化简到阶梯形
8:563×33\times3 矩阵的具体例子
12:19求秩与自由变量个数
15:44题型二:由矩阵结构数维数
17:26零空间很大的情形
21:06平方为零的矩阵可能吗
24:39反例与不可逆性
26:32题型三:由完整解反推矩阵
29:59乘积的零空间
33:27特解加零空间的组合
36:54题型四:真伪判断
40:21可逆矩阵给出的答案
43:47行空间在消元下保持不变
45:33题型六:行空间与零空间的交集

题型一:张成空间的维数

题目:u,v,wu,v,w 是 R7\mathbb{R}^7 中的三个非零向量,它们张成的子空间维数可能是多少?

答案:1、2 或 3。

  • 不超过 3,因为只有三个向量;
  • 不为 0,因为题目说它们非零;
  • 三种都能实现:三个共线则维数 1,共面但不共线则维数 2,不共面则维数 3。

考点是「维数 ≤ 向量个数」以及「非零排除零维」。

题型二:由矩阵结构数维数

题目:UU 是 5×35\times3 的行阶梯形矩阵,秩 r=3r=3。

  1. N(U)N(U) 是什么?
  2. B=[U2U]B=\begin{bmatrix}U\\2U\end{bmatrix} 的秩是多少?
  3. C=[UUU0]C=\begin{bmatrix}U&U\\U&0\end{bmatrix} 的秩是多少?
  4. dim⁡N(C⊤)\dim N(C^\top) 是多少?

解答:

  1. 5×35\times3 且秩 3,三列全部是主元列,没有自由变量,所以 N(U)={0}N(U)=\{\mathbf{0}\}。
  2. 第二块是第一款的两倍,行变换会把它消掉,只剩 UU。秩仍为 3。
  3. CC 是 10×610\times6。对块做消元:用左上块消去左下块,得到 [UU0−U]\begin{bmatrix}U&U\\0&-U\end{bmatrix},再消去右上块,得到 [U00−U]\begin{bmatrix}U&0\\0&-U\end{bmatrix}。两个 UU 各贡献 3 个主元,秩为 6。
  4. C⊤C^\top 是 6×106\times10,有 10 列;秩为 6,所以
dim⁡N(C⊤)=10−6=4.\dim N(C^\top)=10-6=4.

这一组题只用到两条公式:dim⁡N(A)=n−r\dim N(A)=n-r,以及分块消元不改变秩。

题型三:由完整解反推矩阵

题目:已知 Ax=bA\mathbf{x}=\mathbf{b} 的一个特解是 xp=(2,0,0)⊤\mathbf{x}_p=(2,0,0)^\top,零空间的一组基是 (1,1,0)⊤(1,1,0)^\top 与 (0,0,1)⊤(0,0,1)^\top,且 b=(2,4,2)⊤\mathbf{b}=(2,4,2)^\top。求 AA 的秩、列空间与矩阵本身。

第一步:由零空间维数得秩。

零空间维数为 2,矩阵是 3×33\times3,所以

r=n−2=3−2=1.r=n-2=3-2=1.

第二步:由特解得第一列。

AxpA\mathbf{x}_p 只取 AA 的第一列乘 2:

2⋅col1=(2,4,2)⊤⟹col1=(1,2,1)⊤.2\cdot\text{col}_1=(2,4,2)^\top \quad\Longrightarrow\quad \text{col}_1=(1,2,1)^\top.

第三步:由零空间向量得其余列。

(0,0,1)⊤∈N(A)(0,0,1)^\top\in N(A) 说明第三列为零向量。

(1,1,0)⊤∈N(A)(1,1,0)^\top\in N(A) 说明

col1+col2=0⟹col2=(−1,−2,−1)⊤.\text{col}_1+\text{col}_2=\mathbf{0} \quad\Longrightarrow\quad \text{col}_2=(-1,-2,-1)^\top.

于是

A=[1−102−201−10].A=\begin{bmatrix} 1&-1&0\\ 2&-2&0\\ 1&-1&0 \end{bmatrix}.

验证:三列都是 (1,2,1)⊤(1,2,1)^\top 的倍数,秩为 1;列空间是 (1,2,1)⊤(1,2,1)^\top 的所有倍数。因此 Ax=bA\mathbf{x}=\mathbf{b} 有解当且仅当 b\mathbf{b} 是 (1,2,1)⊤(1,2,1)^\top 的倍数。

方法论:完整解 =xp+N(A)=\mathbf{x}_p+N(A) 里包含两类信息——特解给出「某一列与 b\mathbf{b} 的关系」,零空间基给出「列与列之间的线性关系」。两者合起来往往足以还原整个矩阵。

题型四:真伪判断

命题判断理由
方阵 AA 的 N(A)={0}N(A)=\{\mathbf{0}\},则 N(A⊤)={0}N(A^\top)=\{\mathbf{0}\}真方阵满秩,转置也满秩
5×55\times5 可逆矩阵构成子空间假不含零矩阵;两个可逆矩阵之和可能奇异
B2=0⇒B=0B^2=0\Rightarrow B=0假反例 [0100]\begin{bmatrix}0&1\\0&0\end{bmatrix}
m×mm\times m 矩阵列无关,则对任意 b\mathbf{b} 可解真列无关即满秩,方阵满秩即可逆
m=nm=n 时行空间等于列空间假[0100]\begin{bmatrix}0&1\\0&0\end{bmatrix} 的行空间是 (0,1)(0,1) 方向,列空间是 (1,0)(1,0) 方向
AA 与 −A-A 有相同的四个子空间真取负不改变任何线性组合关系
四个子空间相同 ⇒A\Rightarrow A 是 BB 的倍数假任意两个 6×66\times6 可逆矩阵都有相同的四个子空间(全空间与零空间)
交换两行,行空间与零空间不变真行变换保持行空间;方程组的解不变
交换两行,列空间不变假列空间通常改变

为什么「可逆矩阵之和」不一定可逆

取

I+[0100]=[1101]I+\begin{bmatrix}0&1\\0&0\end{bmatrix} =\begin{bmatrix}1&1\\0&1\end{bmatrix}

这个例子仍可逆,但取

[1001]+[000−1]=[1000]\begin{bmatrix}1&0\\0&1\end{bmatrix} + \begin{bmatrix}0&0\\0&-1\end{bmatrix} = \begin{bmatrix}1&0\\0&0\end{bmatrix}

就奇异了。更简单的理由是:可逆矩阵集合不含零矩阵,而子空间必须包含零向量(对任意元素乘 0 仍在集合内),所以它根本不可能是子空间。

题型五:乘积的零空间

题目:B=CDB=CD,其中 CC 可逆。N(B)N(B) 与 N(D)N(D) 有什么关系?

答案:相等。

  • 若 Dx=0D\mathbf{x}=\mathbf{0},则 Bx=C(Dx)=0B\mathbf{x}=C(D\mathbf{x})=\mathbf{0},所以 N(D)⊆N(B)N(D)\subseteq N(B)。
  • 若 Bx=0B\mathbf{x}=\mathbf{0},左乘 C−1C^{-1} 得 Dx=0D\mathbf{x}=\mathbf{0},所以 N(B)⊆N(D)N(B)\subseteq N(D)。

因此左乘一个可逆矩阵不改变零空间。这也解释了为什么消元(左乘初等矩阵的乘积)能用来求零空间。

题型六:行空间与零空间的交集

题目:为什么向量 (1,2,3)(1,2,3) 不可能同时是 AA 的某一行,又属于 N(A)N(A)?

假设它是第一行,且 Av=0A\mathbf{v}=\mathbf{0}。那么 AvA\mathbf{v} 的第一个分量就是

(1,2,3)⋅(1,2,3)=1+4+9=14≠0,(1,2,3)\cdot(1,2,3)=1+4+9=14\neq0,

与 Av=0A\mathbf{v}=\mathbf{0} 矛盾。

更一般地:

行空间∩N(A)={0}.\text{行空间}\cap N(A)=\{\mathbf{0}\}.

若 v\mathbf{v} 既在行空间又在零空间,则 v\mathbf{v} 与行空间中的每个向量都垂直;特别地,它与自己垂直,于是 v⊤v=0\mathbf{v}^\top\mathbf{v}=0,只能是零向量。

这正是下一讲「正交」的入口:行空间与零空间不仅交集为零,而且互相垂直,维数相加为 nn,因此互为正交补。

复习要点

  • 维数计数:dim⁡N(A)=n−r\dim N(A)=n-r,dim⁡N(A⊤)=m−r\dim N(A^\top)=m-r。
  • 分块消元只关心每块贡献多少个主元。
  • 完整解 =xp+N(A)=\mathbf{x}_p+N(A) 可以反推出矩阵的列关系。
  • 子空间必须含零、对加法与数乘封闭;可逆矩阵集合不满足。
  • 左乘可逆矩阵不改变零空间。
  • 行空间与零空间只共享零向量。

理解检验

1. 三个非零向量在 R2\mathbb{R}^2 中能张成几维?为什么?

参考答案:只能张成 1 维或 2 维。

  • 上界:张成空间是 R2\mathbb{R}^2 的子空间,维数不超过 2。
  • 下界:三个向量都非零,张成空间至少含有一个非零方向,维数至少为 1。
  • 两种都能实现:三个向量共线时维数为 1;存在两个不共线的向量时维数为 2。

关键点是:维数不超过向量个数,也不超过所在空间的维数。三个向量在 R2\mathbb{R}^2 中必然相关(3 > 2),但「相关」不影响它们能张成 2 维。

2. 若 UU 是 5×35\times3、秩 2,N(U)N(U) 的维数是多少?

参考答案:用公式 dim⁡N(A)=n−r\dim N(A)=n-r。这里 n=3n=3、r=2r=2,所以

dim⁡N(U)=3−2=1.\dim N(U)=3-2=1 .

从消元看:三列中恰有两个主元列、一个自由列,自由变量只有一个,零空间由对应的一个特殊解张成。(顺带:dim⁡N(U⊤)=m−r=5−2=3\dim N(U^\top)=m-r=5-2=3。)

3. 已知 AA 的零空间基与一个特解,如何确定 AA 的每一列?

参考答案:把「列」当作未知量,两类信息各给出一组方程。

设 AA 的列为 c1,…,cn\mathbf{c}_1,\dots,\mathbf{c}_n。

  • 特解给出:c1xp1+⋯+cnxpn=b\mathbf{c}_1x_{p1}+\cdots+\mathbf{c}_nx_{pn}=\mathbf{b},即列的一个已知组合等于 b\mathbf{b}。
  • 零空间基向量 v\mathbf{v} 给出:c1v1+⋯+cnvn=0\mathbf{c}_1v_1+\cdots+\mathbf{c}_nv_n=\mathbf{0},即列之间的一条齐次线性关系。

共有 1+(n−r)=n−r+11+(n-r)=n-r+1 个向量方程,而未知列有 nn 个;再结合秩为 rr 的条件,通常足以唯一确定各列(若信息不足以定值,则只能确定列空间与列间关系)。本讲例题中,零空间基 e3\mathbf{e}_3 直接说明 c3=0\mathbf{c}_3=\mathbf{0},(1,1,0)(1,1,0) 说明 c1+c2=0\mathbf{c}_1+\mathbf{c}_2=\mathbf{0},特解再定出 c1\mathbf{c}_1,于是整个 AA 被还原。

4. 为什么 m=nm=n 时行空间通常不等于列空间?举一个对称矩阵的例子说明何时相等。

参考答案:行向量与列向量是不同方向上的对象,行空间由行张成、列空间由列张成,二者一般不同。反例:

A=[0100].A=\begin{bmatrix}0&1\\0&0\end{bmatrix}.

它的行空间是 (0,1)(0,1) 方向(第二行非零),列空间是 (1,0)(1,0) 方向(第一列非零),两者是不同的直线。

当 AA 对称时(A=A⊤A=A^\top),行就是列,行空间与列空间自然相等。例如

A=[1223],A=\begin{bmatrix}1&2\\2&3\end{bmatrix},

两行与两列完全相同,行空间 = 列空间 = R2\mathbb{R}^2。更一般地,只要 A⊤A^\top 与 AA 有相同的列空间(例如 AA 对称),两者就相等。

5. 用一句话证明行空间与零空间的交集只有零向量。

参考答案:若 v\mathbf{v} 同时属于行空间与 N(A)N(A),则 Av=0A\mathbf{v}=\mathbf{0} 说明 v\mathbf{v} 与 AA 的每一行都正交,因而与行空间的每个向量正交——特别地与自己正交;于是 v⊤v=0\mathbf{v}^\top\mathbf{v}=0,故 v=0\mathbf{v}=\mathbf{0}。

与前后课程的关系

这一讲把第 1–12 讲压成考点。下一讲从「交集只有零」推进到「互相垂直」,并引出正交补与最小二乘的核心方程 A⊤Ax^=A⊤bA^\top A\hat{\mathbf{x}}=A^\top\mathbf{b}。

What this review covers

The first quiz covers the Ax=bA\mathbf{x}=\mathbf{b} half of the course, with emphasis on rectangular matrices: null spaces, left null spaces, rank, and the four fundamental subspaces. When a matrix is invertible these notions are automatic; when it is rectangular, each one has to be checked.

Below the material is organized by problem type rather than lecture order.

Timeline

TimeTopic
0:09Opening: what this review covers
1:49Type 1: the dimension of a span
5:29Elimination to echelon form
8:56A concrete 3×33\times3 example
12:19Computing rank and free variables
15:44Type 2: counting dimensions from structure
17:26When the null space is large
21:06Can a matrix square to zero
24:39The counterexample and invertibility
26:32Type 3: reconstructing a matrix from its solutions
29:59The null space of a product
33:27Particular solution plus null space
36:54Type 4: true or false
40:21The answer for an invertible matrix
43:47The row space survives row operations
45:33Type 6: row space meets null space

Type 1: dimension of a span

Problem. Let u,v,wu,v,w be nonzero vectors in R7\mathbb{R}^7. What dimensions can their span have?

Answer. 1, 2, or 3.

  • At most 3, since there are only three vectors.
  • Not 0, since they are nonzero.
  • All three occur: collinear gives dimension 1, coplanar but not collinear gives 2, and not coplanar gives 3.

The test is simply "dimension is at most the number of vectors" plus "nonzero rules out zero".

Type 2: counting dimensions from structure

Problem. UU is a 5×35\times3 echelon matrix of rank r=3r=3.

  1. What is N(U)N(U)?
  2. What is the rank of B=[U2U]B=\begin{bmatrix}U\\2U\end{bmatrix}?
  3. What is the rank of C=[UUU0]C=\begin{bmatrix}U&U\\U&0\end{bmatrix}?
  4. What is dim⁡N(C⊤)\dim N(C^\top)?

Solutions.

  1. With three columns and rank 3, every column is a pivot column. No free variables, so N(U)={0}N(U)=\{\mathbf{0}\}.
  2. The second block is twice the first and is eliminated. The rank remains 3.
  3. CC is 10×610\times6. Block elimination gives [U00−U]\begin{bmatrix}U&0\\0&-U\end{bmatrix}; each UU contributes 3 pivots, so the rank is 6.
  4. C⊤C^\top is 6×106\times10 with 10 columns and rank 6:
dim⁡N(C⊤)=10−6=4.\dim N(C^\top)=10-6=4.

Only two facts are used: dim⁡N(A)=n−r\dim N(A)=n-r and block elimination preserves rank.

Type 3: reconstructing a matrix from its complete solution

Problem. For Ax=bA\mathbf{x}=\mathbf{b}, one particular solution is xp=(2,0,0)⊤\mathbf{x}_p=(2,0,0)^\top, a basis of the null space is (1,1,0)⊤(1,1,0)^\top and (0,0,1)⊤(0,0,1)^\top, and b=(2,4,2)⊤\mathbf{b}=(2,4,2)^\top. Find the rank, the column space, and AA itself.

Step 1: rank from the null-space dimension.

The null space has dimension 2 and the matrix is 3×33\times3, so

r=n−2=3−2=1.r=n-2=3-2=1.

Step 2: the first column from the particular solution.

AxpA\mathbf{x}_p is twice the first column:

2⋅col1=(2,4,2)⊤⟹col1=(1,2,1)⊤.2\cdot\text{col}_1=(2,4,2)^\top \quad\Longrightarrow\quad \text{col}_1=(1,2,1)^\top.

Step 3: the remaining columns from the null-space vectors.

(0,0,1)⊤∈N(A)(0,0,1)^\top\in N(A) forces the third column to be zero.

(1,1,0)⊤∈N(A)(1,1,0)^\top\in N(A) gives

col1+col2=0⟹col2=(−1,−2,−1)⊤.\text{col}_1+\text{col}_2=\mathbf{0} \quad\Longrightarrow\quad \text{col}_2=(-1,-2,-1)^\top.

Therefore

A=[1−102−201−10].A=\begin{bmatrix} 1&-1&0\\ 2&-2&0\\ 1&-1&0 \end{bmatrix}.

Check: all three columns are multiples of (1,2,1)⊤(1,2,1)^\top, so the rank is 1 and the column space is the line through (1,2,1)⊤(1,2,1)^\top. The system is solvable exactly when b\mathbf{b} is a multiple of (1,2,1)⊤(1,2,1)^\top.

Method. A complete solution xp+N(A)\mathbf{x}_p+N(A) carries two kinds of information: the particular solution relates one column to b\mathbf{b}, and the null-space basis records linear relations among columns. Together they often determine the whole matrix.

Type 4: true or false

StatementVerdictReason
Square AA with N(A)={0}N(A)=\{\mathbf{0}\} implies N(A⊤)={0}N(A^\top)=\{\mathbf{0}\}TrueFull rank is preserved by transpose
The invertible 5×55\times5 matrices form a subspaceFalseThe zero matrix is missing; sums can be singular
B2=0B^2=0 implies B=0B=0FalseCounterexample [0100]\begin{bmatrix}0&1\\0&0\end{bmatrix}
An m×mm\times m matrix with independent columns is solvable for every b\mathbf{b}TrueIndependent columns in a square matrix mean invertible
When m=nm=n, the row space equals the column spaceFalseFor [0100]\begin{bmatrix}0&1\\0&0\end{bmatrix} the row space is the (0,1)(0,1) direction and the column space the (1,0)(1,0) direction
AA and −A-A have the same four subspacesTrueNegation preserves every linear relation
Same four subspaces implies AA is a multiple of BBFalseAny two invertible 6×66\times6 matrices have the same four subspaces
Swapping two rows leaves the row and null spaces unchangedTrueRow operations preserve the row space; solutions are unchanged
Swapping two rows leaves the column space unchangedFalseThe column space generally changes

Why a sum of invertible matrices can be singular

Take

[1001]+[000−1]=[1000],\begin{bmatrix}1&0\\0&1\end{bmatrix} + \begin{bmatrix}0&0\\0&-1\end{bmatrix} = \begin{bmatrix}1&0\\0&0\end{bmatrix},

which is singular. A simpler argument: the set of invertible matrices does not contain the zero matrix, and a subspace must contain zero, so the set cannot be a subspace at all.

Type 5: null space of a product

Problem. Let B=CDB=CD with CC invertible. How do N(B)N(B) and N(D)N(D) compare?

Answer. They are equal.

  • If Dx=0D\mathbf{x}=\mathbf{0}, then Bx=C(Dx)=0B\mathbf{x}=C(D\mathbf{x})=\mathbf{0}, so N(D)⊆N(B)N(D)\subseteq N(B).
  • If Bx=0B\mathbf{x}=\mathbf{0}, multiply on the left by C−1C^{-1} to get Dx=0D\mathbf{x}=\mathbf{0}, so N(B)⊆N(D)N(B)\subseteq N(D).

Left multiplication by an invertible matrix does not change the null space. This is why elimination, which is left multiplication by a product of elementary matrices, can be used to find null spaces.

Type 6: intersection of row and null spaces

Problem. Why can the vector (1,2,3)(1,2,3) never be both a row of AA and a member of N(A)N(A)?

Suppose it were the first row and Av=0A\mathbf{v}=\mathbf{0}. Then the first component of AvA\mathbf{v} would be

(1,2,3)⋅(1,2,3)=1+4+9=14≠0,(1,2,3)\cdot(1,2,3)=1+4+9=14\neq0,

contradicting Av=0A\mathbf{v}=\mathbf{0}.

More generally,

row space∩N(A)={0}.\text{row space}\cap N(A)=\{\mathbf{0}\}.

If v\mathbf{v} lies in both, it is orthogonal to every vector in the row space, in particular to itself. Then v⊤v=0\mathbf{v}^\top\mathbf{v}=0, so v=0\mathbf{v}=\mathbf{0}.

This is the doorway to the next lecture: the row space and null space are not only disjoint, they are perpendicular, with dimensions adding to nn. They are orthogonal complements.

Review points

  • Dimension counts: dim⁡N(A)=n−r\dim N(A)=n-r and dim⁡N(A⊤)=m−r\dim N(A^\top)=m-r.
  • Block elimination: count pivots contributed by each block.
  • The complete solution xp+N(A)\mathbf{x}_p+N(A) can reveal the columns of AA.
  • A subspace must contain zero and be closed under addition and scaling; invertible matrices satisfy neither.
  • Left multiplication by an invertible matrix preserves the null space.
  • The row space and null space share only the zero vector.

Check your understanding

1. What dimensions can three nonzero vectors span in R2\mathbb{R}^2, and why?

Reference answer: Only dimension 1 or dimension 2.

  • Upper bound: the span is a subspace of R2\mathbb{R}^2, so its dimension is at most 2.
  • Lower bound: all three vectors are nonzero, so the span contains a nonzero direction and has dimension at least 1.
  • Both occur: collinear vectors span a line (dimension 1); two non-collinear vectors among them span the plane (dimension 2).

The key point: the dimension is bounded both by the number of vectors and by the ambient space. Three vectors in R2\mathbb{R}^2 are necessarily dependent (3>23>2), but dependence does not prevent them from spanning dimension 2.

2. If UU is 5×35\times3 with rank 2, what is dim⁡N(U)\dim N(U)?

Reference answer: Use dim⁡N(A)=n−r\dim N(A)=n-r. Here n=3n=3 and r=2r=2, so

dim⁡N(U)=3−2=1.\dim N(U)=3-2=1 .

In elimination terms: among the three columns exactly two are pivots and one is free, so there is a single free variable and the null space is spanned by one special solution. (For reference, dim⁡N(U⊤)=m−r=5−2=3\dim N(U^\top)=m-r=5-2=3.)

3. Given a null-space basis and one particular solution, how do you recover each column of AA?

Reference answer: Treat the columns as unknowns; the two kinds of information supply equations.

Let the columns be c1,…,cn\mathbf{c}_1,\dots,\mathbf{c}_n.

  • The particular solution gives c1xp1+⋯+cnxpn=b\mathbf{c}_1x_{p1}+\cdots+\mathbf{c}_nx_{pn}=\mathbf{b}: one known combination of the columns equals b\mathbf{b}.
  • Each null-space basis vector v\mathbf{v} gives c1v1+⋯+cnvn=0\mathbf{c}_1v_1+\cdots+\mathbf{c}_nv_n=\mathbf{0}: a homogeneous linear relation among the columns.

That is 1+(n−r)1+(n-r) vector equations for nn unknown columns; together with the rank condition they usually determine the columns uniquely (if not, only the column space and the relations are determined). In the lecture's example, the null vector e3\mathbf{e}_3 forces c3=0\mathbf{c}_3=\mathbf{0}, (1,1,0)(1,1,0) forces c1+c2=0\mathbf{c}_1+\mathbf{c}_2=\mathbf{0}, and the particular solution then fixes c1\mathbf{c}_1, recovering all of AA.

4. Why is the row space usually different from the column space when m=nm=n? When are they equal?

Reference answer: Rows and columns are objects in different directions; the row space is spanned by the rows and the column space by the columns, and in general they differ. Counterexample:

A=[0100].A=\begin{bmatrix}0&1\\0&0\end{bmatrix}.

Its row space is the (0,1)(0,1) direction (second row), while its column space is the (1,0)(1,0) direction (first column).

They coincide when AA is symmetric (A=A⊤A=A^\top), since then rows and columns are the same vectors. For example

A=[1223]A=\begin{bmatrix}1&2\\2&3\end{bmatrix}

has identical rows and columns, so both spaces are R2\mathbb{R}^2. More generally, they coincide whenever A⊤A^\top has the same column space as AA.

5. Prove in one sentence that the row space and null space intersect only at zero.

Reference answer: If v\mathbf{v} lies in both, then Av=0A\mathbf{v}=\mathbf{0} makes v\mathbf{v} orthogonal to every row, hence to every vector of the row space — in particular to itself, so v⊤v=0\mathbf{v}^\top\mathbf{v}=0 and therefore v=0\mathbf{v}=\mathbf{0}.

How this connects to the course

This lecture compresses Lectures 1–12 into exam technique. The next moves from "the intersection is only zero" to "they are perpendicular", introducing orthogonal complements and the central least-squares equation A⊤Ax^=A⊤bA^\top A\hat{\mathbf{x}}=A^\top\mathbf{b}.

来源