The Student Room Group

Showing two norms are not equivalent

I am struggling with part d of this question. So the uniform norm is just the supremum ( f(t) ). I tried going for the definition of equivalent norms which I can't seem to find constants to bound each of the norms.

My thought is since the vector space is complete with respect to the uniform norm I could go down the route of showing it is not complete with the a norm(defined in part b). So I would need to show that there's a cauchy sequence that doesn't converge in the space, I am unsure what counterexample I could think of if anybody could point me in a direction to start?

It would be helpful to understand what points I should be looking at and the thought process to constructing a counterexample for such a space

Oh, did not know we had solutions online.. Could not find them on the warwick website at all, thanks!
Original post by VincentCheung
I am struggling with part d of this question. So the uniform norm is just the supremum ( f(t) ). I tried going for the definition of equivalent norms which I can't seem to find constants to bound each of the norms.

My thought is since the vector space is complete with respect to the uniform norm I could go down the route of showing it is not complete with the a norm(defined in part b). So I would need to show that there's a cauchy sequence that doesn't converge in the space, I am unsure what counterexample I could think of if anybody could point me in a direction to start?

It would be helpful to understand what points I should be looking at and the thought process to constructing a counterexample for such a space


faf|f|_a \leq |f|_{\infty} always, since tf(t)f(t)|t f(t)| \leq |f(t)| in this range.

To show inequivalence, then, you need that no constant CC exists with CfafC |f|_a \geq |f|_{\infty} for all ff. Going more concrete, we want a sequence of functions fnf_n with fna1nfn|f_n|_a \geq \frac{1}{n} |f_n|_{\infty}. Can you make that happen?
I understand why we want that but I'm terrible at coming up with an example.. I think the first thing I should do is think of a function so that points fnf_n with fna|f_n|_a goes to 0 but fn|f_n|_{\infty} that doesn't, so I want to essentially fix a value for the uniform norm but let the first one tend to zero?

I have just got the solution so to think of another example would be quite easy but I'm more interested in the thought process in finding the function at this point
Original post by VincentCheung
I understand why we want that but I'm terrible at coming up with an example.. I think the first thing I should do is think of a function so that points fnf_n with fna|f_n|_a goes to 0 but fn|f_n|_{\infty} that doesn't, so I want to essentially fix a value for the uniform norm but let the first one tend to zero?

Yep, that's one way to do it. What might a function look like if its a-norm is very small but its sup is high?
The solution said 1-tn for 0 <= t <= 1/n
and 0 otherwise, so I'm trying to look past that

would e^-n work? Since the uniform norm is 1 and the a norm is 0 at this same point
Original post by VincentCheung
The solution said 1-tn for 0 <= t <= 1/n
and 0 otherwise, so I'm trying to look past that

would e^-n work? Since the uniform norm is 1 and the a norm is 0 at this same point

The function fn(x)=enf_n(x) = e^{-n} is constant - is that really what you meant?
Sorry,

e^-nx is what I meant, I think this works?
Original post by VincentCheung
Sorry,

e^-nx is what I meant, I think this works?

Yep, that works. You just need any function which starts off nonzero but then shrinks to zero very rapidly.
Ahh perfect, I have a slightly better idea now. Thanks for your help!

Quick Reply

Latest