# 2 definitions and a theorem relating them

1. Oct 21, 2012

### dextercioby

1. The problem statement, all variables and given/known data

Let E be a vector space and p,q 2 norms on it. By definition p,q induce the same topology on E, iff they assign the same neighborhood basis to the 0 vector. *QUESTION: What does the bolded part mean ? Does this mean that, if whatever A included in the system (=basis?) induced by p, A is included in a certain B in the system induced by q and viceversa, i.e. any B in the system induced by q is included in a certain A induced by p ? If so, then see my arguments below.

The 2nd definition would be:

E a vector space and p, q norms on it. p, q are said to be equivalent iff there exist c, C>0 such as

$$c p(x) \leq q(x) \leq C p(x), ~\forall x \in E$$

The theorem would be the equivalence of the 2 definitions, namely

Theorem: p,q are equivalent iff they induce the same topology on E.

The problem would be to prove the theorem.

3. The attempt at a solution

Assume p, q equivalent. Then if $\epsilon >0$, $\{ x\in E, p(x) \leq \epsilon/c \}$ is a n-hood system of 0 wrt p. Let A be one of the n-hoods in this system. Then $A\subseteq \{x\in E, q(x) \leq\epsilon \}$ ?? That would mean that $\{x\in E, q(x) \leq\epsilon \}$ is also a n-hood system of 0 induced by q. Using now the 2nd inequality would mean

Let $\epsilon>0$. Then [itex] \{x\in E, q(x)\leq \epsilon/C \} is a n-hood system of 0 induced by q and let B be a n-hood from it. Does it mean (using the 2nd inequality in the definition of equivalence) that:

$$B\subseteq \{x\in E, p(x)\leq \epsilon\}$$ ??

If so, then the 2 systems of n-hoods would be the same (??) so that p,q induce the same topology.

This would be half the proof. How's the other half ?

Last edited: Oct 21, 2012