So, my question is, where did I go wrong in my approach?

Doom of Doom
Messages
85
Reaction score
0
Let S be a subspace of L^{2}(\left[0,1\right]) and suppose \left|f(x)\right|\leq K \left\| f \right\| for all f in S.

Show that the dimension of S is at most K^{2}

---------

The prof hinted us to use Bessel's inequality.

Namely, let \left\{ u_1,\dots, u_m \right\} be a set of orthonormal vectors in L^{2}(\left[0,1\right]). Then \left\| f \right\|^2 \geq \sum_{k=1}^m \left| \left\langle f, u_k \right\rangle\right|^2

I just keep getting stuck, and getting things like

\left\|f \right\|^2 =\int_0^1 \left|f(x)\right|^2 dx \leq \int_0^1 K^2 \left\|f \right\|^2 dx = K^2 \left\|f \right\|^2

and I can't figure out how to apply Bessel's inequality.

I guess the goal is to assume m linear independent vectors, and show that m is less than K^2. Help, please?
 
Physics news on Phys.org
haven't totally put it together yet, but I'm thinking start with assuming an orthonormal basis for S:{...,ui ...} of dimension m.

Try expanding f in terms of the basis function & then knowing both the ui and f satsify the given inequality, hopefully you can find a contradiction if m>K2 as you say
 
Last edited:
considering a function which is a certain linear combination of an orthogonal basis for S should help. As S is a subspace, its closed under linear combinations, so the function is contained in S
 
Last edited:
Still at a loss...

So: Assume {u1, ..., um} is an orthonormal basis of S.
I'm trying the function f= u1 + ... + um. Is this the right one to consider? It certainly is the nicest.

Then ||f|| = sqrt(m).

I still don't know where exactly to apply the Bessels inequality. If I do, all I keep getting is that
m^2 < m^2 K
which tells me nothing.

More help? Thanks for the response.
 
Last edited:
must be missing something.. thought it would follow on, but when i went to work it completely, went in a circle...

now I'm not even so sure i understand the question correctly, what about sinusoidal basis functions, couldn't you have an infinite orthogonal number of those with ||f|| =1, and maximum magnitude ~ sqrt(2)

will pass it on to some of the other guys to have a look
 
Last edited:
Ok, I'm started off with a different approach to see if I could understand this a bit better... and it only leads me to believe that the problem must be wrong.
----

Assume K=1 (since that is the smallest it can be). Then S is the set of functions on [0,1] such that |f(x)| < ||f|| for all x in [0,1]. For simplicity, let's just say that ||f|| =1.

But the integral of |f(x)|^2 from 0 to 1 must be 1, and |f(x)| can never be less than 1. Thus, it must be the case that |f(x)|=1 for all x on the interval.

So the modulus of f(x) is always 1, and thus f(x)=e^{ia(x)} for some function a(x).

Then the inner product of f with some other function g is
\left\langle f,g\right\rangle=\int_0^1 e^{i(a(x)-b(x))}dx.

But this integral does not necessarily have to have a modulus of 1 (indeed consider a(x)=2x and b(x)=x), and thus g is not a multiple of f. This leads me to believe that there must be more than one function in the basis for S, a contradiction to the dimension of S being less than or equal to K=1.
 
Last edited:
Prove $$\int\limits_0^{\sqrt2/4}\frac{1}{\sqrt{x-x^2}}\arcsin\sqrt{\frac{(x-1)\left(x-1+x\sqrt{9-16x}\right)}{1-2x}} \, \mathrm dx = \frac{\pi^2}{8}.$$ Let $$I = \int\limits_0^{\sqrt 2 / 4}\frac{1}{\sqrt{x-x^2}}\arcsin\sqrt{\frac{(x-1)\left(x-1+x\sqrt{9-16x}\right)}{1-2x}} \, \mathrm dx. \tag{1}$$ The representation integral of ##\arcsin## is $$\arcsin u = \int\limits_{0}^{1} \frac{\mathrm dt}{\sqrt{1-t^2}}, \qquad 0 \leqslant u \leqslant 1.$$ Plugging identity above into ##(1)## with ##u...
Back
Top