Showing a sequence of functions is Cauchy/not Cauchy in L1

  • #1
Euler2718
90
3

Homework Statement



Determine whether or not the following sequences of real valued functions are Cauchy in [itex]L^{1}[0,1][/itex]:

(a) [tex] f_{n}(x) = \begin{cases} \frac{1}{\sqrt{x}} & , \frac{1}{n+1}\leq x \leq 1 \\ 0 & , \text{ otherwise } \end{cases} [/tex]

(b) [tex]
f_{n}(x) = \begin{cases} \frac{1}{x} & , \frac{1}{n+1}\leq x < 1 \\ 0 & , \text{ otherwise } \end{cases}
[/tex]

Homework Equations



[itex]\{ f_{n} \}_{n=1}^{\infty}[/itex] is Cauchy in [itex]L^{1}[0,1][/itex] iff for all [itex]\epsilon>0[/itex] there exists [itex]N\in\mathbb{N}[/itex] such that for [itex]n,m\geq N[/itex], [itex]||f_{n}-f_{m}||_{1} < \epsilon[/itex].

The Attempt at a Solution


[/B]
In both problems I think I should get to a step where I integrate [itex]\left|\frac{1}{\sqrt{x}}\right| [/itex] and [itex]\left|\frac{1}{x}\right| [/itex] over [itex][0,1][/itex] and get the norm values of [itex]2[/itex] and [itex]undefined[/itex] respectively, then I can conclude easily. Embarrassingly I don't know what [itex]f_{n}-f_{m}[/itex] is explicitly (in terms of its piecewise definition). I had first thought it to be zero for both cases, I don't think this is the case. Feel stupid asking this, but how do you subtract [itex]f_{n}[/itex] and [itex]f_{m}[/itex] in either case ?
 

Answers and Replies

  • #2
Ray Vickson
Science Advisor
Homework Helper
Dearly Missed
10,706
1,722

Homework Statement



Determine whether or not the following sequences of real valued functions are Cauchy in [itex]L^{1}[0,1][/itex]:

(a) [tex] f_{n}(x) = \begin{cases} \frac{1}{\sqrt{x}} & , \frac{1}{n+1}\leq x \leq 1 \\ 0 & , \text{ otherwise } \end{cases} [/tex]

(b) [tex]
f_{n}(x) = \begin{cases} \frac{1}{x} & , \frac{1}{n+1}\leq x < 1 \\ 0 & , \text{ otherwise } \end{cases}
[/tex]

Homework Equations



[itex]\{ f_{n} \}_{n=1}^{\infty}[/itex] is Cauchy in [itex]L^{1}[0,1][/itex] iff for all [itex]\epsilon>0[/itex] there exists [itex]N\in\mathbb{N}[/itex] such that for [itex]n,m\geq N[/itex], [itex]||f_{n}-f_{m}||_{1} < \epsilon[/itex].

The Attempt at a Solution


[/B]
In both problems I think I should get to a step where I integrate [itex]\left|\frac{1}{\sqrt{x}}\right| [/itex] and [itex]\left|\frac{1}{x}\right| [/itex] over [itex][0,1][/itex] and get the norm values of [itex]2[/itex] and [itex]undefined[/itex] respectively, then I can conclude easily. Embarrassingly I don't know what [itex]f_{n}-f_{m}[/itex] is explicitly (in terms of its piecewise definition). I had first thought it to be zero for both cases, I don't think this is the case. Feel stupid asking this, but how do you subtract [itex]f_{n}[/itex] and [itex]f_{m}[/itex] in either case ?

If ##m > n##, in what ##x##-region are ##f_m(x)## and ##f_n(x)## the same? In what ##x##-region are they different? So, what is the formula for ##f_m(x) - f_n(x)## over ##0 \leq x \leq 1?##
 
  • #3
pasmith
Homework Helper
2022 Award
2,584
1,183
Your functions are defined piecewise, and vanish on [itex][0, \frac{1}{n+1})[/itex]. You shouldn't be getting any undefined integrals:
[tex]
\int_0^1 f_n(x)\,dx = \int_0^{1/(n+1)} 0\,dx + \int_{1/(n+1)}^1 \frac1x\,dx = \left[\log(x)\right]_{1/(n+1)}^1 = -\log(1/(n+1)),
[/tex] etc.

You can without loss of generality assume [itex]N \leq n < m[/itex].
 
  • #4
Euler2718
90
3
If ##m > n##, in what ##x##-region are ##f_m(x)## and ##f_n(x)## the same? In what ##x##-region are they different? So, what is the formula for ##f_m(x) - f_n(x)## over ##0 \leq x \leq 1?##

Using (a) as an example, if [itex]m>n[/itex], then [itex]f_{m}[/itex] and [itex]f_{n}[/itex] would be the same (both [itex]\frac{1}{\sqrt{x}}[/itex] ) on [itex]\frac{1}{n+1} \leq x \leq 1[/itex] and different ([itex]f_{m}=\frac{1}{\sqrt{x}}[/itex] but [itex]f_{n}=0[/itex]) on [itex] \frac{1}{m+1}\leq x < \frac{1}{n+1}[/itex]. So [itex] \displaystyle f_{m}-f_{n} = \begin{cases} \frac{1}{\sqrt{x}} &, \frac{1}{m+1}\leq x < \frac{1}{n+1} \\ 0 &, \text{ otherwise } \end{cases} [/itex] ?

Your functions are defined piecewise, and vanish on [itex][0, \frac{1}{n+1})[/itex]. You shouldn't be getting any undefined integrals:
[tex]
\int_0^1 f_n(x)\,dx = \int_0^{1/(n+1)} 0\,dx + \int_{1/(n+1)}^1 \frac1x\,dx = \left[\log(x)\right]_{1/(n+1)}^1 = -\log(1/(n+1)),
[/tex] etc.

You can without loss of generality assume [itex]N \leq n < m[/itex].

I see what you mean. I was not careful enough with the permissible [itex]x[/itex] values to notice this.
 

Suggested for: Showing a sequence of functions is Cauchy/not Cauchy in L1

  • Last Post
Replies
6
Views
632
Replies
8
Views
1K
Replies
7
Views
430
Replies
19
Views
642
Replies
13
Views
470
  • Last Post
Replies
3
Views
317
Replies
11
Views
435
Top