Are the Metrics \rho^{(p)} and \rho^{(q)} Equivalent on \Re^n?

  • Thread starter Thread starter jjou
  • Start date Start date
  • Tags Tags
    Equivalence
jjou
Messages
63
Reaction score
0
Show equivalence of family of metrics on \Re^n: \rho^{(p)}:(x,y)\rightleftharpoons(\sum_{i=1}^{n}|x_i-y_i|^p)^{\frac{1}{p}} for p\geq1
The attempt at a solution
Want to show for p,q\geq1, p\neq q, that \rho^{(p)} and \rho^{(q)} generate the same topology. I tried two methods:

1. Show that a set is open in (\Re^n,\rho^{(p)}) iff it is open in (\Re^n,\rho^{(q)}).

2. Show that the bases are equivalent (an element of the base of the topology generated by \rho^{(p)} is a union of elements of the base of the topology generated by \rho^{(q)}.

Both reduced down to considering, for a fixed x_0\in\Re^n and r_p, the set B_{r_p}(x_0)=\{x\in\Re^n|\rho^{(p)}(x,x_0)<r_p\}. Then fix a point y\in B_{r_p}(x_0) and show there exists a r_q such that:

for any u satisfying \rho^{(q)}(u,y)<r_q, u also satisfies \rho^{(p)}(u,y)<r_p-\rho^{(p)}(y,x_0). Ie. that any point in B_{r_q}(y) is also in B_{r_p}(x_0).

Also tried to get ideas by simplifying problem to \Re^2 with p=1 and p=2. Was not very illuminating since I was still stuck working with the same expressions.

----

Another method is to show that for any x,y\in\Re^n there exists c,C>0 such that:
c\rho^{(p)}(x,y)\leq\rho^{(q)}(x,y)\leq C\rho^{(p)}(x,y)
Ie. that:
c(\sum_{i=1}^{n}|x_i-y_i|^p)^{\frac{1}{p}}\leq(\sum_{i=1}^{n}|x_i-y_i|^q)^{\frac{1}{q}}\leq C(\sum_{i=1}^{n}|x_i-y_i|^p)^{\frac{1}{p}}

I didn't know whether to attempt to show both inequalities at once or separately, but either way, I'm not sure how to manipulate all the exponents and absolute values...

Please help!
 
Physics news on Phys.org
Each of these metrics comes from a norm, and as such, they are all equivalent because R^n is finite dimensional.

If you don't know this, then I suggest you prove each of the metrics is equivalent to \rho^{(1)}. Holder's inequality will be useful here, because we can write (for p>1):

\sum |x_i - y_i| \leq \left(\sum |x_i - y_i|^p\right)^{1/p} \left(\sum 1^{p/(p-1)}\right)^{(p-1)/p}
 
Thank you so much! Very simple solution using Holder's, if I'm not wrong...

Using c(\sum_{i=1}^{n}|x_i-y_i|^p)^{\frac{1}{p}}\leq(\sum_{i=1}^{n}|x_i-y_i|^q)^{\frac{1}{q}}\leq C(\sum_{i=1}^{n}|x_i-y_i|^p)^{\frac{1}{p}}, take
c=1 and C=\left(\sum 1^{p/(p-1)}\right)^{(p-1)/p}=n^{(p-1)/p}.

The first inequality holds iff c^p(\sum_{i=1}^{n}|x_i-y_i|^p)\leq(\sum_{i=1}^{n}|x_i-y_i|)^p which is obviously true by expansion.

The second inequality fulfills Holder's.

Yes?

Again, thanks so much! :)
 
Looks good (as long as that 'q' is actually a '1'!). :smile:
 
There are two things I don't understand about this problem. First, when finding the nth root of a number, there should in theory be n solutions. However, the formula produces n+1 roots. Here is how. The first root is simply ##\left(r\right)^{\left(\frac{1}{n}\right)}##. Then you multiply this first root by n additional expressions given by the formula, as you go through k=0,1,...n-1. So you end up with n+1 roots, which cannot be correct. Let me illustrate what I mean. For this...
Back
Top