Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $X$ be a normed vector space, $x\in X$ and $Z\subseteq X$. Then we define the point-to-set distance function as: $$ \|x\|_Z = \inf_{z\in Z} \| x-z\| $$ I use the notation $\|\cdot\|_Z$ for convenience without implying that the operation is a norm. Assuming that $Z$ is convex we may not conclude that: $$ \|x+y\|_Z \leq \|x\|_Z + \|y\|_Z $$ What sort of properties should $Z$ possess so that the triangle inequality holds for all $x\in X$? For example, $Z$ being a singleton brings about the desired inequality but I'm looking for a less trivial example.

Additionally, let $A\in M_n(\mathbb{R})$ be a matrix and let $Z\neq X$. I noticed that we may define:

$$ \|A\|_Z:= \sup_{x\in\mathbb{R}^n,x\notin Z}\frac{\|Ax\|_Z}{\|x\|_Z} $$

while $\|A\|_X:=0$.

Then, it should be true that: $$ \|Ax\|_Z \leq \|A\|_Z\cdot\|x\|_Z $$

So, this is one norm-like property for $\|\cdot\|_Z$.

What other properties of $\|\cdot\|_Z$ can we deduce based on assumed properties of $Z$?

Update: As Davide Giraudo pointed out, the triangle inequality holds true in case $Z$ is a linear subspace of $X$. It seems that it holds for convex cones too (for nonconvex cones it's easy to find a counter example on the plane). Let us assume that $Z$ is a convex cone. Then:

$$ \|x+y\|_Z=\inf_{z\in Z}\|x+y-z\|\leq \|x+y-(z_1+z_2)\| $$

where $z_1,z_2\in Z$ and we proceed as in David's answer. Here, we used the fact that whenever $Z$ is a convex cone, then $z_1,z_2\in Z \Rightarrow z_1+z_2\in Z$ so any vector in $Z$ can be decomposed into a sum of vectors of $Z$ and vice versa.

Question: Would be nice to identify other classes of sets for which the induced point-to-set distance satisfies the triangle inequality.

share|cite|improve this question

If $Z$ is a vector subspace then for $x,y\in X$, $z_1,z_2\in Z$ we have $$||x+y||_Z\leq ||x+y-(z_1+z_2)||\leq ||x-z_1||+||y-z_2|||,$$ and taking the infimum with respect to $z_1$ we get: $||x+y||_Z\leq ||x||_Z+||y-z_2||$. Now taking the infimum with respect to $z_2$ we get $||x+y||_Z\leq ||x||_Z+||y||_Z$.

In fact when $Y$ is a closed subspace of $X$, we can consider the equivalence relation $x\sim y$ iff $x-y\in Y$ and denoting $[x]$ the class of $X$, we can put $||[x]||:= ||x||_Y$. It defines a norm on $X/Y$ and their are many links between the properties of $Y$, $X$, $X/Y$, named "three spaces property". Namely, consider a property (P), like separability, completeness, etc... The three spaces property means that if between $X,Y,X/Y$ two have the property (P) then the third will also have the property (P).

share|cite|improve this answer
I have the feeling that the triangle inequality also holds for cases in which $Z$ is a cone... – Pantelis Sopasakis Feb 9 '12 at 22:05
Yes, you're right. – Davide Giraudo Feb 9 '12 at 22:06

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.