找回密码
 快速注册
搜索
查看: 70|回复: 11

two isomorphic $C(ℝ)$-modules whose presentation matrices are not equivalent

[复制链接]

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65397
QQ

显示全部楼层

hbghlyj 发表于 2023-3-4 07:20 |阅读模式
Show that if $R$ is the ring of continuous functions on $ℝ$ with pointwise addition and multiplication then there are $A, B ∈ M_1(R)$ with $A ≁ B$, and for which $R^1/R^1A$ is $R$-linearly isomorphic to $R^1/R^1B$.



Related:
Determining the presentation matrix for a module
$\ker_F (R.) \cong \ker_F (R'.)$ does not imply: $R$ and $R'$ are equivalent matrices

48

主题

992

回帖

1万

积分

积分
14981
QQ

显示全部楼层

Czhang271828 发表于 2023-3-4 13:47
本帖最后由 Czhang271828 于 2023-3-6 22:37 编辑 Proposition 1 Let $R$ be any commutative ring. $I$ and $J$ are ideals of $R$. If $R/I\simeq R/J$ as $R$-modules with the standard $R$-module structure, then $I=J$. In other words, isomorphic modules have the same annihilators.

Proof of Prop 1 For isomorphic $R$-modules $\varphi: R/I\simeq R/J$ and arbitrary $x\in J$, we have that
$$
\varphi (x+I)=x\varphi(1+I)=x(1+J)=0+J.
$$
Thus $x+I\in \mathrm{ker}(\varphi)$. Since $\varphi $ is isomorphism, we have that $x\in I$. It yields that $J\subseteq I$. Similarly, $I\subseteq J$, thus we have $I=J$.

Proposition 2 Let $R$ be any commutative ring. $I$ and $J$ are ideals of $R$. The ring isomorphism $\varphi :R/I\simeq R/J$ no longer implies $I\simeq J$. Here $\varphi$ preserves $\{+,\cdot ,0,1\}$. In other words, isomorphic quotient rings have the isomorphic annihilators.

Proof of Prop 2 A counterexample: (1) set $R:=\mathcal C^0(\mathbb R,\mathbb R)$, $R/(x)\simeq R/(x-1)$ by translation. (2) For any commutative ring $A$, $(A\times A)/(0\times A)\simeq (A\times A)/(A\times 0)$.

Lemma 0 This lemma may help with your initial question. Set $R:=\mathcal C^0(\mathbb R,\mathbb R)$. If we have principal ideals $(f)=(g)$, then there exists invertible $h\in R$ such that $fh=g$.

Proof of Lemma 0 There exsits $h,h'\in R$ such that $fh=g$ and $gh'=f$. Since $fhh'=f$, $gh'h=g$, we have that $hh'=1$ on $\overline{\mathrm{supp}(f)}=\overline{\mathrm{supp}(g)}$, which is the disjoint union of closed intervals.

Set $\overline{\mathrm{supp}(f)}=\mathbb R\setminus (0,1)$ without the loss of generality. Since $h$ and $h'$ are uniformly continuous on $[-1,0]$ and $[1,2]$, they are bounded. Let $h_-$ (resp. $h_+$) be the restriction of $h$ on $(-\infty,0]$ (resp. $[1,+\infty)$), so is $h'_-$ (resp. $h'_+$). We proved that
$$
S=\{h_-(0),h_-'(0),h_+(1),h_+'(1)\}
$$
is bounded. Since $h_-(0)h'_-(0)=h_+(1)h'_+(1)=1$​, $0\notin S$​. Define
$$
\tilde h(x)=\left\{\begin{align*}
&|h(x)|&&x\in \overline{\mathrm{supp}(f)},\\
&|h(0)|(1-x)+|h(1)|x&&x\in(0,1).
\end{align*}\right.
$$
Set $\tilde h'(x)=1/\tilde h(x)$. Here $\tilde h(x)$ and $\tilde h'(x)$ are non-vanishing and thus invertible. Then we have either
$$
f\tilde h=g\quad g\tilde h'=f,
$$
or
$$
(f-2f_+)\tilde h=(g-2g_+)\quad (g-2g_+)\tilde h'=(f-2f_+).
$$
Here we have the same ideals $(f)=(f-2f_+)$ and $(g)=(g-2g_+)$.

The hidden post may contain ambiguous expression, factual errors, or dogmatic opinions.
**Question.** Set $R:=$ the ring of continuous functions from $\mathbb R$ to $\mathbb R$.
We say $f,g\in R$ are equivalent, whenever there exists invertible $h_1,h_2\in R$ such that $h_1 fh_2=g$.
Let $(\tau)$ be the principal ideal generated by $\tau\in R$.
We shall prove that $R/(f)$ and $R/(g)$ are isomorphic $R$-modules whenever $f$ and $g$ are almost equivalent.

**Proof.**  When $f$ and $g$ are equivalent, there exists invertible $r\in R$ such that $f=gr$. Then we have the following isomorphism
$$
\varphi :R/(f)\to R/(g)\quad \overline{h}\mapsto \overline{hr}.
$$
To see this,

1. $\varphi(\overline 0)=\varphi(\overline {f})=\overline{fr}=\overline g=\overline 0$, thus $\varphi$ preserves $0$.
2. $\varphi$ preserves the structure of additive group.
3. $\varphi (h\cdot \overline p)=\overline {hpr}=h\overline {pr}=h\cdot \varphi(\overline p)$.

Clearly, $\varphi$ is an homomorphism with inverse
$$
\varphi^{-1} :R/(g)\to R/(f)\quad \overline{h}\mapsto \overline{hr^{-1}}.
$$
Thus is an isomorphism.

We say $f$ and $g$ are almost equivalent, the sets of zeros differ on finite points. Here we only need to prove $R/(x)\simeq R/(1)$ and use induction. Define
$$
\varphi:R/(1)\to R/(x),\quad \overline{f(x)-f(\epsilon)}\mapsto \overline{f(x)-x\epsilon^{-1}f(\epsilon)}.
$$
Here we fix $\epsilon$ such that $f(\epsilon )\neq 0\neq g(\epsilon)$. We see $\varphi$ is a bijection, since both $f(x)-f(\epsilon)$ and $f(x)-x\epsilon^{-1}f(\epsilon)$ are unique representative element in the corresponding equivalence classes.

The converse part is similar.

Here a new question is raised, do we have $\overline{\mathrm{supp}}(f)=\overline{\mathrm{supp}}(g)$ whenever $R/(f)=R/(g)$? I believe the answer is Yes.

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65397
QQ

显示全部楼层

 楼主| hbghlyj 发表于 2023-3-4 22:19
本帖最后由 hbghlyj 于 2023-3-4 23:16 编辑
Czhang271828 发表于 2023-3-4 06:47
definitions of $R'$, $M_1$?

$R^1$ is the free module of rank 1 over $R$ (it is just $R$).
$M_1(R)$ is $1\times1$ matrices over $R$ (it is just $R$).
There's no $R'$ in the problem.

About $R'$ in that PDF: I searched up that PDF on Google, but it's not a part of the problem, though it seems to be related

48

主题

992

回帖

1万

积分

积分
14981
QQ

显示全部楼层

Czhang271828 发表于 2023-3-4 23:22
hbghlyj 发表于 2023-3-4 22:19
$R^1$ is the free module of rank 1 over $R$ (it is just $R$).
$M_1$ is $1\times1$ matrices over $R$ ...


$A\sim B$ means $A=B$. For nonzero $A$, $A\not\sim 2A$, but $R/AR=R/2AR$.  

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65397
QQ

显示全部楼层

 楼主| hbghlyj 发表于 2023-3-5 03:52

According to Wikipedia, $A\sim B$ doesn't imply $A=B$. It means $\operatorname{rank}A=\operatorname{rank}B$.
en.wikipedia.org/wiki/Matrix_equivalence
Matrix equivalence is an equivalence relation on the space of rectangular matrices.
For two rectangular matrices of the same size, their equivalence can also be characterized by the following conditions
$\bullet\;$The matrices can be transformed into one another by a combination of elementary row and column operations.
$\bullet\;$Two matrices are equivalent if and only if they have the same rank.

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65397
QQ

显示全部楼层

 楼主| hbghlyj 发表于 2023-3-5 06:15
Czhang271828 发表于 2023-3-4 16:22
For nonzero $A$, $A\not\sim 2A$, but $R/AR=R/2AR$.


$A=(2I)A$, so $A\sim2A$

48

主题

992

回帖

1万

积分

积分
14981
QQ

显示全部楼层

Czhang271828 发表于 2023-3-5 13:59
hbghlyj 发表于 2023-3-5 03:52
According to Wikipedia, $A\sim B$ doesn't imply $A=B$. It means $\operatorname{rank}A=\operatorname ...


You should first clarify what $\sim$ stands for, since $\sim$ means equivalence relation of similar matrices in most of the time.

Back to your question, $A\sim B$ whenever there exists invertible $C$ and $D$ such that $CAD=B$. In $\mathcal C(\mathbb R)$ sense, $C$ and $D$ are continuous functions without zeros. Set $A:f(x)\mapsto x\cdot f(x)$, $B:f(x)\mapsto (x-1)f(x)$. Then $A\not \sim B$. However, $R/AR$ is $\mathbb R$-linearly isomorphic to $R/BR$ by translating $\overline{f(t)}\mapsto \overline{f(t-1)}$.

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65397
QQ

显示全部楼层

 楼主| hbghlyj 发表于 2023-3-5 19:40
Czhang271828 发表于 2023-3-5 06:59
However, $R/AR$ is $\mathbb R$-linearly isomorphic to $R/BR$ by translating $\overline{f(t)}\mapsto \overline{f(t-1)}$.


I don't think $\phi:\overline{f(t)}\mapsto \overline{f(t-1)}$ is $R$-linear?
$\phi((x+1)\times (x+1))=x^2$
$(x+1)\times\phi(x+1)=x(x+1)$
But $x^2+R^1B\ne x(x+1)+R^1B$
since $x^2|_{x=1}=1\ne 2=x(x+1)|_{x=1}$

48

主题

992

回帖

1万

积分

积分
14981
QQ

显示全部楼层

Czhang271828 发表于 2023-3-5 22:25
hbghlyj 发表于 2023-3-5 19:40
I don't think $\phi:\overline{f(t)}\mapsto \overline{f(t-1)}$ is $R$-linear?
$\phi((x+1)\ti ...

The question is restated and answered in $\#2$.

3149

主题

8387

回帖

6万

积分

$\style{scale:11;fill:#eff}꩜$

积分
65397
QQ

显示全部楼层

 楼主| hbghlyj 发表于 2023-3-6 18:48
Sorry, I read your post but still can't find $A, B ∈ M_1(R)$ that satisfy the conditions in the question

点评

See $\#2$ again.  发表于 2023-3-6 20:46

手机版|悠闲数学娱乐论坛(第3版)

GMT+8, 2025-3-4 18:07

Powered by Discuz!

× 快速回复 返回顶部 返回列表