Forgot password?
 Create new account
View 428|Reply 8

rank(AᵀA)=rank(A)与长方形矩阵的逆

[Copy link]

3151

Threads

8498

Posts

610K

Credits

Credits
66208
QQ

Show all posts

hbghlyj Posted at 2022-1-6 11:44:17 |Read mode
Last edited by hbghlyj at 2023-1-3 23:47:00$A$是$n$阶方阵,$A'$是$A$的转置,$A'A= I$,求证:$AA'= I$

3151

Threads

8498

Posts

610K

Credits

Credits
66208
QQ

Show all posts

 Author| hbghlyj Posted at 2022-1-21 02:52:16
Last edited by hbghlyj at 2023-1-3 14:34:00回复 3# kuing
$n=\operatorname{rank}I=\operatorname{rank}(A'A)≤\operatorname{rank}A⇒A$可逆
由$A'A=I$得$A'=A^{-1}$,所以$AA'=I$.

3151

Threads

8498

Posts

610K

Credits

Credits
66208
QQ

Show all posts

 Author| hbghlyj Posted at 2022-1-21 03:41:10
Last edited by hbghlyj at 2022-1-24 06:32:00把"方阵"的条件去掉就不成立了:

➢ $A$是$m×n$阶矩阵,$A'$是$A$的转置,$A'A= I_n$,则$AA'= I_m$.


例如$A=\left(
\begin{array}{ccc}
1 & 0  \\
0 & 1  \\
0 & 0
\end{array}
\right)$,$A'A=I_2$,而$AA'=\left(
\begin{array}{ccc}
1 & 0 & 0 \\
0 & 1 & 0 \\
0 & 0 & 0 \\
\end{array}
\right)\ne I_3$.
几何上,$A$是$\mathbb R^2\to\mathbb R^3$的正交变换.

48

Threads

969

Posts

110K

Credits

Credits
14870
QQ

Show all posts

Czhang271828 Posted at 2022-1-23 15:55:07
回复 2# hbghlyj



其实, 实数域上 $\mathrm{rank}(A^TA)=\mathrm{rank}(A)$.

一方面, $Ax=0$ 能推出 $A^TAx=0$.

另一方面, $A^TAx=0$ 能推出 $x^TA^TAx=0$, 等价于 $Ax=0$.

复数就把转置变成共轭转置.

3151

Threads

8498

Posts

610K

Credits

Credits
66208
QQ

Show all posts

 Author| hbghlyj Posted at 2023-1-3 21:24:02

略作补充:

Czhang271828 发表于 2022-1-23 08:55
回复 2# hbghlyj
其实, 实数域上 $\mathrm{rank}(A^TA)=\mathrm{rank}(A)$.
一方面, $Ax=0$ 能推出 $A^TAx=0$.
另一方面, $A^TAx=0$ 能推出 $x^TA^TAx=0$, 等价于 $Ax=0$.
复数就把转置变成共轭转置.

以上证明的$Ax = 0 ⇔ A^T Ax = 0$推出$N(A) = N(A^T A)$. 由$\dim N(A) +\operatorname{rank}(A) = n$知$$\operatorname{rank}(A^TA) =\operatorname{rank}(A)$$
把$A$换成$A^T$得$$\operatorname{rank}(AA^T) =\operatorname{rank}(A^T)$$
又$\operatorname{rank}(A)=\operatorname{rank}(A^T)$就有$$\operatorname{rank}(A^TA)=\operatorname{rank}(AA^T)$$

3151

Threads

8498

Posts

610K

Credits

Credits
66208
QQ

Show all posts

 Author| hbghlyj Posted at 2023-1-3 22:30:37
hbghlyj 发表于 2022-1-20 20:41
把"方阵"的条件去掉就不成立了:
➢ $A$是$m×n$阶矩阵,$A'$是$A$的转置,$A'A= I_n$,则$AA'= I_m$.
例如$A=\ ...

Can the product of two rectangular matrices be an identity matrix?

对于矩阵$A$, 存在矩阵$B$使$BA=I$的充要条件是$N(A)=0$.
证明
[$f\circ g=\text{id}$则$g$为单射]由$BA=I$推出$N(A)=0$.
反过来,若$N(A)=0$,
5#证明了$N(A)=N(A^\top A)$,
且$A^\top A$是方阵, 所以$A^\top A$可逆, 则$B=(A^\top A)^{-1}A^\top$满足$BA=I$.
满足$BA=I$的矩阵$B$是不唯一的.

搬运Quora用的代码
document.querySelectorAll('script[type="math/tex"]').forEach(a=>a.parentNode.replaceWith('$'+a.innerText+'$'))

3151

Threads

8498

Posts

610K

Credits

Credits
66208
QQ

Show all posts

 Author| hbghlyj Posted at 2023-1-3 22:43:55
hbghlyj 发表于 2023-1-3 15:30
满足$BA=I$的矩阵$B$是不唯一的.

$\{B\in M_{n\times m}\mid BA=I\}$的$n(m-n)$维仿射空间. 见楼下.

48

Threads

969

Posts

110K

Credits

Credits
14870
QQ

Show all posts

Czhang271828 Posted at 2023-1-3 23:03:39
Last edited by Czhang271828 at 2023-1-3 23:10:00
hbghlyj 发表于 2023-1-3 22:43
求$\{B\in M_{n\times m}\mid BA=1\}$的维数?
这不是线性空间啊, 何来维数?

如果 $m\geq n$, $A\in M_{m\times n}$ 是给定的秩 $n$ 矩阵, 则 $\{v\in M_{m\times 1} \mid v^TA=O\}$ 是 $m-n$ 维空间, 从而 $\{B\in M_{n\times m}\mid BA=O\}$ 是 $n(m-n)$ 维空间.  

Comment

哦对! 已修改  Posted at 2023-1-4 05:15

手机版Mobile version|Leisure Math Forum

2025-4-21 14:25 GMT+8

Powered by Discuz!

× Quick Reply To Top Return to the list