4

$(I)$

I've been browsing some problems concerning metrics not induced by norms, and I've found a comment that said that such a metric should be a concave monotone function. Here is the post I'm referring to.

Could you tell me why that is?

$(II)$

I know that if a metric $d: X \times X \rightarrow \mathbb{R}$ ($X$ is a vector space with scalars in $\mathbb{K}$ ) satisfies:

$(1) d(\lambda x, \lambda y) = |\lambda| d(x,y) \ \ \ \forall x,y \in X, \ \lambda \in \mathbb{K}$, in particular

$d(\lambda x,0) = |\lambda| d(x,0) = |\lambda| \ ||x||$

$(2) d(x+w, y+w) = d(x,y) \ \ \forall x,y,w \in X$, in particular

$d(x-y,0)=d(x,y)$

then the function $\|u\|:= d(u,0)$ is a norm. I have problems proving the converse, meaning that if a metric defines a norm, then it satisfies the two conditions $(1), \ (2)$.

Could you help me with that, too?

Is it true to say that if we want a metric to be induced by a norm it should satisfy both conditions mentioned above?

I would really appreciate all your insight.

Community
  • 1
Bilbo
  • 1,323
  • 3
    I'm not entirely sure what you mean here. As written here, it sounds like you have problems proving that the metric induced by a norm satisfies $(1)$ and $(2)$ (which is the easy direction), but I think when you say "the converse" you are referring to "Conversely, if a metric has the above properties, then $d(u,0)$ is a norm." from the answer you linked to. This makes your question very confusing. – kahen Oct 01 '13 at 09:25
  • 1
    Ok. My question is could you help me prove that if a metric defines a norm, then it satisfies the two conditions. The second question is what condition should a metric satisfy in order for it not to be induced by any metric. – Bilbo Oct 01 '13 at 09:35
  • 2
    "what condition should a metric satisfy in order for it not to be induced by any metric" Huh ?? – bubba Oct 01 '13 at 11:09
  • 2
    You may ((1)trim down to $d(\lambda x,0)=|\lambda|x$ and (2)to $d(x-y,0)=d(x,y)$, it's more intuitive. – Michael Hoppe Oct 01 '13 at 13:31
  • 1
    See also http://math.stackexchange.com/questions/166380/not-every-metric-is-induced-from-a-norm – Martin Sleziak Oct 24 '14 at 08:46

1 Answers1

3

such a metric should be a concave monotone function

Nobody actually said that in the linked post. A metric is a function on $X\times X$, where $X$ is the underlying set of points. Being concave and monotone is a property that a function on $\mathbb R$ might have, not a function on $X\times X$.

What is true: if $d$ is a metric and $\phi:[0,\infty)\to [0,\infty)$ is an increasing concave function such that $\phi(0)=0$, then $\phi\circ d$ is also a metric.

I have problems proving the converse, meaning that if a metric defines a norm, then it satisfies the two conditions (1), (2).

So, the assumption is that $d(x,y)=\|x-y\|$ where $\|\cdot\|$ is a norm. Just go through the list, replacing $d(x,y)$ with $\|x-y\|$ everywhere, and checking that equality holds by virtue of the norm properties. For example, $$d(\lambda x, \lambda y) = \|\lambda x-\lambda y\| = |\lambda| \,\|x-y\| = |\lambda|\,d( x, y)$$ and so forth.