seminorm (original) (raw)
Let V be a real, or a complex vector space, with K denoting the corresponding field of scalars. A seminorm is a function![]()
from V to the set of non-negative real numbers, that satisfies the following two properties.
| p(k𝐮) | =|k|p(𝐮),k∈K,𝐮∈V | Homogeneity | | --------- | ----------------------- | ------------ | | p(𝐮+𝐯) | ≤p(𝐮)+p(𝐯),𝐮,𝐯∈V, | Sublinearity |
A seminorm differs from a norm in that it is permitted that p(𝐮)=0for some non-zero 𝐮∈V.
It is possible to characterize the seminorms properties geometrically. For k>0, let
denote the ball of radius k. The homogeneity property is equivalent![]()
to the assertion that
in the sense that𝐮∈B1 if and only ifk𝐮∈Bk.Thus, we see that a seminorm is fully determined by its unit ball. Indeed, given B⊂V we may define a functionpB:V→ℝ+by
| pB(𝐮)=inf{λ∈ℝ+:λ-1𝐮∈B}. |
|---|
The geometric nature of the unit ball is described by the following.
Proposition 1
The function pB satisfies the homegeneity property if and only if for every u∈V, there exists a k∈R+∪{∞} such that
| λ𝐮∈B if and only if ∥λ∥≤k. |
|---|
Proof. First, let us suppose that the seminorm is both sublinear and homogeneous, and prove that B1 is necessarily convex. Let𝐮,𝐯∈B1, and let k be a real number between 0 and 1. We must show that the weighted average k𝐮+(1-k)𝐯 is in B1 as well. By assumption,
| p(k𝐮+(1-k)𝐯)≤kp(𝐮)+(1-k)p(𝐯). |
|---|
The right side is a weighted average of two numbers between 0 and 1, and is therefore between 0 and 1 itself. Therefore
as desired.
Conversely, suppose that the seminorm function is homogeneous, and that the unit ball is convex. Let 𝐮,𝐯∈V be given, and let us show that
The essential complication here is that we do not exclude the possibility that p(𝐮)=0, but that 𝐮≠0. First, let us consider the case where
By homogeneity, for every k>0 we have
and hence
as well. By homogeneity, again,
Since the above is true for all positive k, we infer that
as desired.
Next suppose that p(𝐮)=0, but that p(𝐯)≠0. We will show that in this case, necessarily,
Owing to the homogeneity assumption, we may without loss of generality assume that
For every k such that 0≤k<1 we have
| k𝐮+k𝐯=(1-k)k𝐮1-k+k𝐯. |
|---|
The right-side expression is an element of B1 because
Hence
and since this holds for k arbitrarily close to 1 we conclude that
The same argument![]()
also shows that
| p(𝐯)=p(-𝐮+(𝐮+𝐯))≤p(𝐮+𝐯), |
|---|
and hence
as desired.
Finally, suppose that neither p(𝐮) nor p(𝐯) is zero. Hence,
are both in B1, and hence
| p(u)p(u)+p(v)𝐮p(u)+p(v)p(u)+p(v)𝐯p(v)=𝐮+𝐯p(u)+p(v) |
|---|
is in B1 also. Using homogeneity, we conclude that
as desired.