24.12.2012 Views

Basic Analysis – Gently Done Topological Vector Spaces

Basic Analysis – Gently Done Topological Vector Spaces

Basic Analysis – Gently Done Topological Vector Spaces

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

46 <strong>Basic</strong> <strong>Analysis</strong><br />

Proposition 5.14 Suppose that ℓ is a linear functional on the vector space X<br />

such that ℓ does not vanish on the whole of X. Then kerℓ is a maximal proper<br />

subspace. Conversely, any maximal proper subspace of X is the kernel of a linear<br />

functional.<br />

Proof It is clear that the kernel of any linear functional ℓ is a linear subspace of<br />

X. Moreover, kerℓ is a proper subspace if and only if ℓ is not identically zero.<br />

Suppose that ℓ(z) �= 0 for some z ∈ X. For any x ∈ X, we have<br />

x = ℓ(x)<br />

�<br />

z + x−<br />

ℓ(z) ℓ(x)<br />

ℓ(z) z<br />

�<br />

.<br />

But x−(ℓ(x)/ℓ(z))z ∈ kerℓ, so it follows that kerℓ is maximal.<br />

Now suppose that V is a maximal proper subspace of X, and let z be any<br />

element of X such that z /∈ V. Since V is maximal, any x ∈ X can be written as<br />

x = αz +v, for suitable α ∈ K and v ∈ V. Moreover, this decomposition of x is<br />

unique; ifalsox = α ′ z+v ′ forsomeα ′ ∈ Kandv ′ ∈ V, then0 = (α−α ′ )z+(v−v ′ ).<br />

Since z /∈ V, we must have α = α ′ and therefore v = v ′ . Define ℓ : X → K by<br />

ℓ(x) = α, where x = αz+v, as above, Evidently, ℓ is well-defined and is linear. If<br />

x = αz +v and ℓ(x) = 0, we have α = 0 and so x = v ∈ V. Clearly, ℓ(v) = 0 for<br />

v ∈ V and so kerℓ = V.<br />

Proposition 5.15 Let ℓ 1 and ℓ 2 be linear functionals on the vector space X,<br />

neither being identically zero. Then ℓ 1 and ℓ 2 have the same kernel if and only if<br />

they are proportional.<br />

Proof Suppose that kerℓ 1 = kerℓ 2 . Let z ∈ X with ℓ 1 (z) �= 0. For any x ∈ X,<br />

x−(ℓ 1 (x)/ℓ 1 (z))z ∈ kerℓ 1 = kerℓ 2 so that ℓ 2 (x) = ℓ 1 (x)ℓ 2 (z)/ℓ 1 (z). Thus ℓ 1 and<br />

ℓ 2 are proportional.<br />

Clearly, if ℓ 1 and ℓ 2 are proportional, then they have the same kernel since the<br />

constant of proportionality is not zero, by hypothesis.<br />

Proposition 5.16 Let ℓ and ℓ 1 ,...ℓ n be linear functionals on the vector space<br />

X. Then either<br />

(i) ℓ is a linear combination of ℓ 1 ,...ℓ n , or<br />

(ii) there is z ∈ X such that ℓ(z) = 1 and ℓ 1 (z) = ··· = ℓ n (z) = 0.<br />

Proof By considering a subset of the ℓ i if necessary, we may assume, without<br />

loss of generality, that ℓ 1 ,...ℓ n are linearly independent. Define the map γ : X →<br />

K n+1 by<br />

γ(x) = (ℓ(x),ℓ 1 (x),...,ℓ n (x)), x ∈ X.<br />

Department of Mathematics King’s College, London

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!