3

$\mathbf {The \ Problem \ is}:$ Consider $\mathbb R^n$ with the norms $N_p(x)=\|x\|_p = [\sum_{j=1}^n |x_j|^p]^{1/p} .$

Show that if there is a linear isometry $\phi : (\mathbb R^n,N_p) \to (\mathbb R^n,N_q)$ and $q \neq p,$ then $(1/p + 1/q) =1.$

$\mathbf {My \ approach}:$ I am trying to use the linearity of $\phi$ and if $\{e_i | I=1(|)n\}$ is the standard basis of $\mathbb R^n$, then we can see some values of $\|\phi(e_i)\|_q =1$ for all $i$ ;

But ,I am getting stuck at proving the relation between $p$ and $q$ .

A small hint is very much required at this moment, thanks in advance .

  • I'm stuck too. Maybe there's some mileage in that all norms are equivalent in a finite dimensional space ? – Tom Collinge Mar 20 '21 at 23:06
  • https://math.stackexchange.com/questions/617942/ellp-is-not-isometric-to-ellq - someone suggested me this answer – Rabi Kumar Chakraborty Mar 21 '21 at 06:17
  • Looking at some of these links I notice when some negative proofs are given there is a reliance on properties only applicable in infinite dimensional cases. So this finite case can still be true. Where did you see it, or is it your own line of enquiry ? – Tom Collinge Mar 21 '21 at 08:17
  • I saw it in a question paper of examination of Multivariate Calculus course – Rabi Kumar Chakraborty Mar 21 '21 at 13:36

0 Answers0