Proving the L¹ norm is a norm and completeness of C⁰([-1,1])

  • Thread starter Thread starter Neoma
  • Start date Start date
Join the discussion
Registration is free. Start your own thread to ask a follow-up.
3 replies · 4K views
Neoma
Messages
9
Reaction score
0
We consider the space [tex]C^0 ([-1,1])[/tex] of continuous functions from [tex][-1,1][/tex] to [tex]\mathbb{R}[/tex] supplied with the following norm:

[tex]||f||_1 = \int_{-1}^{1} |f(x)| dx[/tex]

a. Show that [tex]||.||_1[/tex] defines indeed a norm.

b. Show that the sequence of functions [tex](f_n)[/tex], where

[tex] \begin{align*}<br /> f_n(x) &= -1, \quad & -1 \leq{x} \leq{\frac{-1}{n}} \\<br /> \ &= nx, \quad & \frac{-1}{n} \leq{x} \leq{\frac{1}{n}} \\<br /> \ &= 1, \quad & \frac{1}{n} \leq{x} \leq{1}<br /> \end{align*}[/tex]

is a Cauchy-sequence with respect to the given norm.

c. Show that [tex]C^0 ([-1,1])[/tex] is not complete with respect to the given norm.

I figured out a. myself, by showing this norm satisfies the properties of a norm, but I can't find out how to tackle b. and c.
 
Physics news on Phys.org
b) I assume you can compute [tex]||f_m - f_n||_1[/tex] for arbitrary m > n > 0. Now [tex]f_n[/tex] is called a Cauchy-sequence if for all [tex]\epsilon >0[/tex] there exists a N>0 such that: [tex]||f_m - f_n||_1 < \epsilon[/tex] for all m>n>N. You can do the estimates yourself.

c) Does the sequence converge to a continuous function? A set with a given norm is called closed if all Cauchy-sequences converge to an element in the same set.
 
I knew the definition of a Cauchy sequence, but I still can't find the solution.

Timbuqtu said:
b) I assume you can compute [tex]||f_m - f_n||_1[/tex] for arbitrary m > n > 0.
I think I can't, given a certain n and [tex]f_n[/tex] I can find [tex]f_{n+1}[/tex] and I see that once n approaches infinity [tex]f_n[/tex] becomes either -1 or 1, but I don't know how to work from there. In fact this is all quite new to me.
 
When we assume m > n:

[tex]\begin{align*}|f_m(x) -f_n(x)| &= 0 & x > 1/n \\<br /> \ &= 1-nx & 1/m < x < 1/n \\<br /> \ &= (m-n)x & 0 \leq x < 1/m \end{align*}[/tex]

and [tex]|f_m(-x) -f_n(-x)| = |f_m(x) -f_n(x)|[/tex]. So:

[tex]||f_m - f_n||_1 = \int_{-1}^{1} |f_m(x)-f_n(x)| dx = 2 \int_{0}^{1} |f_m(x)-f_n(x)| dx = 2 ( \int_{0}^{1/m} (m-n)x dx + \int_{1/m}^{1/n} (1-n x) dx ) =[/tex]
[tex]= 2 ( \frac{m-n}{2 m^2} + 1/n - 1/m - \frac{n}{2 n^2} + \frac{n}{2 m^2}) = 1/n-1/m < 1/n[/tex]

Let [tex]\epsilon > 0[/tex]. Take [tex]N > 1/\epsilon[/tex], then for all n,m > N, we have [tex]||f_m - f_n||_1 < 1/N < \epsilon[/tex].

Now we have proven that [tex]f_n[/tex] is indeed a Cauchy-sequence.