Proving that $X$ is a Banach Space and $Y$ is Not

  • Context: MHB 
  • Thread starter Thread starter mathmari
  • Start date Start date
  • Tags Tags
    Banach Space
Click For Summary

Discussion Overview

The discussion centers on proving that the space of $T$-periodic continuous functions, denoted as $X$, is a Banach space, while the space of $T$-periodic continuously differentiable functions, denoted as $Y$, is not. The conversation involves exploring the completeness of these spaces under the maximum norm and examining specific sequences to demonstrate the properties of completeness.

Discussion Character

  • Technical explanation
  • Mathematical reasoning
  • Debate/contested

Main Points Raised

  • Some participants propose that $X$ is complete because every Cauchy sequence in $X$ converges to a limit in $X$.
  • Others argue that $Y$ is not complete, as demonstrated by the sequence $f_n(x)=\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}$, which is Cauchy in $Y$ but does not converge to a limit in $Y$.
  • Participants discuss the necessity of showing that $C(\mathbb{R})$ is a Banach space and question the implications of the limit of a sequence of functions in $C(\mathbb{R})$.
  • There is a suggestion that the Cauchy sequence should be in $Y$, but its limit is not in $Y$, raising concerns about the completeness of $Y$.
  • One participant provides a detailed argument showing that the sequence is uniformly Cauchy, thus reinforcing the claim that $Y$ is not complete.

Areas of Agreement / Disagreement

Participants generally agree that $X$ is a Banach space and that $Y$ is not complete. However, there are ongoing discussions about the justification of certain properties and the implications of the sequences involved, indicating some areas of uncertainty and exploration.

Contextual Notes

Some participants express confusion regarding the relationship between the completeness of $C(\mathbb{R})$ and the properties of the sequences in $Y$. There are also discussions about the conditions under which the limit of a sequence in $C(\mathbb{R})$ remains in that space.

Who May Find This Useful

This discussion may be useful for students and researchers interested in functional analysis, particularly those studying properties of Banach spaces and the completeness of function spaces.

mathmari
Gold Member
MHB
Messages
4,984
Reaction score
7
Hey! :o

Let $ T> 0 $ be fixed.

We denote $ X = \{f \in C (\mathbb{R}): f (t) = f (t + T) \ \forall t \in \mathbb {R} \} $ and $ Y = \{f \in C^1 (\mathbb{R}): f (t) = f (t + T) \ \forall t \in \mathbb {R} \} $ be the spaces of the $ T $ periodic continuous and continuously differentiable functions respectively, both given the maximum norm $ \displaystyle {\| f \|_{\infty} = \max_{t \in \mathbb {R}} | f (t) |} $.

I want to show that $X$ is a Banach space, but not $Y$ as a subspace of $X$.

Hint: Show that $Y$ is not complete. For that, show that the sequence $f_n(x)=\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}$ is a Cauchy sequence in $Y$ but doesn't converge in $Y$.
First we have to show that $X$ is a Banach space, i.e., that $X$ is complete, or not?

It holds that $C(\mathbb{R})$ is a Banach space, or not? Then $f_n(x)\in X\subseteq C(\mathbb{R})$. So a Cauchy sequence $\{f_n\}$ of $C(\mathbb{R})$ converges to $f\in C(\mathbb{R})$. It is left to show that if $f_n$ is $T$-periodic then $f$ is $T$-periodic.
Suppose that $f$ is not $T$-periodic, then $f(t)\neq f(t+T)$ for $t\in \mathbb{R}$.
Since $f_n$ is $T$ periodic, we have that $f_n(t)=f_n(t+T)$. We take the limit $n\rightarrow\infty$ and get $f(t)=f(t+T)$, a contradiction.
That means that a Cauchy sequence $\{f_n\}$ of $X$ converges to $X$, and so $X$ is complete.

Is this correct? (Wondering) Then we want to show that $Y$ is not a Banach space, don't we?

Let $N\in \mathbb{N}$ and $n,m>N$.

We have to show that $$|f_n(x)-f_m(x)|=\left| \frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}-\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{m}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}\right |<\epsilon$$ for every $\epsilon>0$ or not?

Could you give me a hint how we could show that? (Wondering) Then we have that $$\lim_{n\rightarrow \infty}f_n(x)=\lim_{n\rightarrow \infty}\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}=\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\left |\sin\left (\frac{2\pi}{T}x\right )\right |}=\frac{\left |\sin\left (\frac{2\pi}{T}x\right )\right |^2}{\left |\sin\left (\frac{2\pi}{T}x\right )\right |}=\left |\sin\left (\frac{2\pi}{T}x\right )\right |$$ Since the absolute value is not continuously differentiable in $\mathbb{R}$, since it is not differentiable at $0$, it follows that the limit of $f_n(x)$ is not in $Y$ and so $f_n(x)$ doesn't convereg in $Y$.

So, since not every Cauchy sequence of $Y$ is convergent in $Y$, it follows that $Y$ is not complete, right? (Wondering)
 
Last edited by a moderator:
Physics news on Phys.org
Hey mathmari! (Wave)

What you have is correct.
So $X$ is indeed complete and therefore a Banach space.
And $Y$ has a Cauchy sequence that shows it is not complete.
Good job! (Happy)
 
I like Serena said:
What you have is correct.
So $X$ is indeed complete and therefore a Banach space.
And $Y$ has a Cauchy sequence that shows it is not complete.
Good job! (Happy)

Great! (Yes)

How could we justify that $C(\mathbb{R})$ is a Banach space? (Wondering)

Do we have to show that $\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}$ is a Cauchy sequence of $Y$ ?
If yes, how could we do that? (Wondering)
 
mathmari said:
How could we justify that $C(\mathbb{R})$ is a Banach space?

$C(\mathbb{R})$ means that $\forall t\in \mathbb R: \lim_{h\to 0} f(t+h)=f(t)$.
Will that be true if $f$ is the absolute converging limit of a sequence of functions in $C(\mathbb R)$? (Wondering)

mathmari said:
Do we have to show that $\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}$ is a Cauchy sequence of $Y$ ?

That's not related to $C(\mathbb R)$ is it?
Didn't you already show that that particular Cauchy sequence is not in $Y$, showing that $Y$ is not a Banach space? (Worried)
 
I like Serena said:
$C(\mathbb{R})$ means that $\forall t\in \mathbb R: \lim_{h\to 0} f(t+h)=f(t)$.
Will that be true if $f$ is the absolute converging limit of a sequence of functions in $C(\mathbb R)$? (Wondering)

I got stuck right now. I haven't really understood that part. (Thinking)
I like Serena said:
That's not related to $C(\mathbb R)$ is it?
Didn't you already show that that particular Cauchy sequence is not in $Y$, showing that $Y$ is not a Banach space? (Worried)

I thought that the Cauchy sequence should be in $Y$ but its limit is not in $Y$. (Wondering)
 
mathmari said:
Do we have to show that $\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}$ is a Cauchy sequence of $Y$ ?
If yes, how could we do that? (Wondering)
$$|f_n(x)-f_m(x)|=\left| \frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}-\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{m}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}\right | = \left| \frac{\left(\frac1m - \frac1n\right)\sin^2\left (\frac{2\pi}{T}x\right )}{\left(\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right)\right|\right) \left(\frac{1}{m}+\left |\sin\left (\frac{2\pi}{T}x\right)\right|\right) }\right| \leqslant \left| \frac{\left(\frac1m - \frac1n\right)\sin^2\left (\frac{2\pi}{T}x\right )}{\left |\sin\left (\frac{2\pi}{T}x\right)\right|^2}\right| = \left|\tfrac1m - \tfrac1n\right|,$$ which can be made less than $\varepsilon$ for all sufficiently large $m$ and $n$, because the sequence $\left\{\frac1n\right\}$ is Cauchy in $\Bbb{R}$. What's more, the right side of the inequality $|f_n(x)-f_m(x)| < \left|\tfrac1m - \tfrac1n\right|$ is independent of $x$. That shows that the sequence $\{f_n(x)\}$ is uniformly Cauchy, in other words it is Cauchy for the $\|\,.\|_\infty$-norm. That is the norm in both $Y$ and $X$. So $\{f_n(x)\}$ is Cauchy in both $Y$ and $X$.

Since $X$ is complete, it follows that the sequence $\{f_n(x)\}$ converges to a limit in $X$. You have shown that the sequence converges (pointwise) to a non-differentiable function in $X$. But if a sequence has a limit in the $\|\,.\|_\infty$-norm, then that limit must coincide with the pointwise limit. Since that limit is not differentiable, it is not in $Y$. So the sequence does not converge in $Y$ and therefore $Y$ is not complete.
 
Opalg said:
$$|f_n(x)-f_m(x)|=\left| \frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}-\frac{\sin^2\left (\frac{2\pi}{T}x\right )}{\frac{1}{m}+\left |\sin\left (\frac{2\pi}{T}x\right )\right |}\right | = \left| \frac{\left(\frac1m - \frac1n\right)\sin^2\left (\frac{2\pi}{T}x\right )}{\left(\frac{1}{n}+\left |\sin\left (\frac{2\pi}{T}x\right)\right|\right) \left(\frac{1}{m}+\left |\sin\left (\frac{2\pi}{T}x\right)\right|\right) }\right| \leqslant \left| \frac{\left(\frac1m - \frac1n\right)\sin^2\left (\frac{2\pi}{T}x\right )}{\left |\sin\left (\frac{2\pi}{T}x\right)\right|^2}\right| = \left|\tfrac1m - \tfrac1n\right|,$$ which can be made less than $\varepsilon$ for all sufficiently large $m$ and $n$, because the sequence $\left\{\frac1n\right\}$ is Cauchy in $\Bbb{R}$. What's more, the right side of the inequality $|f_n(x)-f_m(x)| < \left|\tfrac1m - \tfrac1n\right|$ is independent of $x$. That shows that the sequence $\{f_n(x)\}$ is uniformly Cauchy, in other words it is Cauchy for the $\|\,.\|_\infty$-norm. That is the norm in both $Y$ and $X$. So $\{f_n(x)\}$ is Cauchy in both $Y$ and $X$.

Since $X$ is complete, it follows that the sequence $\{f_n(x)\}$ converges to a limit in $X$. You have shown that the sequence converges (pointwise) to a non-differentiable function in $X$. But if a sequence has a limit in the $\|\,.\|_\infty$-norm, then that limit must coincide with the pointwise limit. Since that limit is not differentiable, it is not in $Y$. So the sequence does not converge in $Y$ and therefore $Y$ is not complete.

I understand! Thank you very much! (Smile)
 

Similar threads

  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 21 ·
Replies
21
Views
3K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 11 ·
Replies
11
Views
3K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 16 ·
Replies
16
Views
2K
  • · Replies 11 ·
Replies
11
Views
2K