Minkowski Functional of Balanced Convex Absorbing Set in Vector Space is Seminorm
Theorem
Let $\GF \in \set {\R, \C}$.
Let $X$ be a vector space over $\GF$.
Let $A \subseteq X$ be a set that is balanced, convex and absorbing.
Let $\mu_A$ be the Minkowski functional of $A$.
Then $\mu_A$ is a seminorm.
Proof
Case 1: $\GF = \R$
From Balanced Set in Vector Space is Symmetric, $A$ is symmetric.
Hence by Minkowski Functional of Symmetric Convex Absorbing Set in Real Vector Space is Seminorm, $\mu_A$ is a seminorm in this case.
$\Box$
Case 2: $\GF = \C$
From Minkowski Functional of Convex Absorbing Set is Positive Homogeneous, $\mu_A$ is a sublinear functional.
Hence we have:
- $\map {\mu_A} {x + y} \le \map {\mu_A} x + \map {\mu_A} y$ for all $x, y \in X$
and hence Seminorm Axiom $\text N 3$: Triangle Inequality.
We also have:
- $\map {\mu_A} {r x} = r \map {\mu_A} x$ for all $r \in \hointr 0 \infty$ and $x \in X$.
Towards Seminorm Axiom $\text N 2$: Positive Homogeneity, we want to show that:
- $\map {\mu_A} {r x} = \map {\mu_A} x$ for all $r \in \C$ with $\cmod r = 1$.
From the definition of a balanced set, we have:
- $r A \subseteq A$
and since $\cmod {r^{-1} } = 1$, we have:
- $r^{-1} A \subseteq A$
Then we have:
- $r A \subseteq A \subseteq r A$
so that:
- $A = r A$
Then we have:
\(\ds \map {\mu_A} {r x}\) | \(=\) | \(\ds \inf \set {t > 0 : r x \in t A}\) | Definition of Minkowski Functional | |||||||||||
\(\ds \) | \(=\) | \(\ds \inf \set {t > 0 : x \in t r^{-1} A}\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \inf \set {t > 0 : x \in t A}\) | since $\cmod {r^{-1} } = 1$ | |||||||||||
\(\ds \) | \(=\) | \(\ds \map {\mu_A} x\) | Definition of Minkowski Functional |
Now take $r \in \C \setminus \hointl 0 \infty$ and $x \in X$.
We have:
\(\ds \map {\mu_A} {r x}\) | \(=\) | \(\ds \map {\mu_A} {\frac r {\cmod r} \cmod r x}\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \cmod r \map {\mu_A} {\frac r {\cmod r} x}\) | Definition of Sublinear Functional | |||||||||||
\(\ds \) | \(=\) | \(\ds \cmod r \map {\mu_A} x\) | since $\cmod {\dfrac r {\cmod r} } = 1$ |
proving Seminorm Axiom $\text N 2$: Positive Homogeneity.
$\blacksquare$