Linear Algebra ReSupply Date X.

Dual Space

Definition. Dual Space

Let V be a vector space over F , then dual space of V is defined by the vector space L(V, F ).
We denoted by V ∗ .

Remark.
By above definition, we have

(a) V ∗ is a vector space over F .

(b) If dim(V ) < ∞, then we have

dim(V ∗ ) = dim(L(V, F )) = dim(V ) · dim(F ) = dim(V ) · 1 = dim(V )

(c) V ∗∗ is the dual space of V ∗ .

Concept.
If dim(V ) < ∞, then we find a natural identification of V and V ∗∗ .

Theorem.
Let V be a vector space with dim(V ) = n < ∞, and let β = {x1 , x2 , ..., xn } be a ordered basis
for V , then we introduce fi : V → F defined by fi (xj ) = δij i.e.

for all x ∈ V we have [x] = (x1 , x2 , ..., xn )T such that fi (x) = ai for all i

Now, we let β ∗ = {f1 , f2 , ..., fn },then β ∗ is a ordered basis for V ∗ and for all f ∈ V ∗ , we have

X
n
f = f (x1 )f1 + f (x2 )f2 + · · · + f (xn )fn = f (xi )fi
i=1

Proof.
Let f ∈ V ∗ be given, and let dim(V ∗ ) = n, then we need to show

X
n
f = f (x1 )f1 + f (x2 )f2 + · · · + f (xn )fn = f (xi )fi
i=1

(Then β ∗Pis a basis.)

Let g = ni=1 f (xi )fi , then we have

X
n X
n
g(xj ) = f (xi )fi (xj ) = f (xi )δij = f (xj )
i=1 i=1

for all j ∈ [1, n] ∩ N, and hence f ≡ g.

1 | Linear Algebra ReSupply Date X JunWei,Zhang Edit

 Definition. Dual Basis
Let V be a vector space with dim(V ) = n < ∞, and let β = {x1 , x2 , ..., xn } be a ordered basis
for V , and we introduce fi : V → F defined by fi (xj ) = δij .
Then β ∗ = {f1 , f2 , ..., fn } is called the dual basis of V

Definition.
Let V, W be two vector space with ordered basis β, γ and dim(V ), dim(W ) < ∞, and let T ∈
L(V, W ), then we defined the transpose of T by T t : W ∗ → V ∗ .
In particular, g : W → F ∈ W ∗ → gT ∈ V ∗ .

Theorem. ([FIS] theorem 2.25)

Let V, W be two vector space with ordered basis β, γ and dim(V ), dim(W ) < ∞, and let T ∈
L(V, W ), then we have

(a) T t ∈ L(W ∗ , V ∗ ).
∗
(b) [T ]βγ ∗ = ([T ]βγ )T .

Proof.
Let β = {x1 , x2 , ..., xn }, γ = {y1 , y2 , ..., ym }, and let

β ∗ = {f1 , f2 , f3 , ..., fn−1 , fn }, γ ∗ = {g1 , g2 , g3 , ..., gm−1 , gm }

Then we show that the following:

(i) We show that linear:

Let h1 , h2 ∈ W ∗ be given, and let c ∈ F be given, then we have

T t (ch1 + h2 ) = (ch1 + h2 )T = ch1 T + h2 T = cT t (h1 ) + T t (h2 )

(ii) We show that transpose matrix: ∗

We consider that the j-th column of [T t ]βγ ∗ i.e. we consider that
X
T t (gj ) = gj T = (gj T )(xs )fs
s

∗
Thus, we know that the row i, column j of [T t ]βγ ∗ is

X
m X
m X
m
(gj T )(xi ) = gj (T (xi )) = gj ( Aki yk ) = Aki gj (yk ) = Aki · δjk = Aji
k=1 k=1 k=1

∗
where A = [T ]γβ , and hence [T ]βγ ∗ = ([T ]βγ )T .

2 | Linear Algebra ReSupply Date X JunWei,Zhang Edit

 Definition.
b : V ∗ → F i.e.b
Let x ∈ V be given, then we defined by x x(f ) = f (x).

Theorem.
Let x ∈ V be given, then x
b is linear.

Proof.
By definition, we have

b(af + g) = (af + g)(x) = af (x) + g(x) = ab

x b(g)
x(f ) + x

for all a ∈ F , f, g ∈ V ∗ .

Lemma 1.
b(f ) = 0 for all f ∈ V ∗ , then
Let V be a vector space over F with dim(V ) < ∞, and let x ∈ V . If x
x = 0V .

Proof.
If x ̸= 0V , then we choose an ordered basis

β = {x1 , x2 , ..., xn } & β ∗ = {f1 , f2 , ..., fn }

such that
b(f1 ) = f1 (x) = 1
x

Theorem. ([FIS] theorem 2.26)

Let V be a vector space over F woth dim(V ) < ∞. If ψ : V → V ∗∗ defined by ψ(x) = x
b, then ψ
is isomorphic.

Proof.
We show that the following:

(i) We show that ψ ∈ L(V ∗∗ , F ):

Let x, y ∈ V be given, and let c ∈ F be given, then we have

\
ψ(cx + y) = cx x(f ) + gb = cψ(x) + ψ(y)
+ y(f ) = f (cx + y) = cf (x) + f (y) = cb

for all f ∈ V ∗ .

3 | Linear Algebra ReSupply Date X JunWei,Zhang Edit

 Proof. ([FIS] theorem 2.26)
We show that the following:

(ii) We show that ψ is one-to-one:

If ψ(x) = 0 ∈ V ∗∗ i.e. ψ(x) = x
b = 0 ∈ V ∗∗ , then we have x
b(f ) = 0 for all f ∈ V ∗ .
By Lemma, we get x = 0V , and hence ker(ψ) = {0V }.
Thus, we know that ψ is one-to-one.

Since dim(V ) = dim(V ∗ ∗), then we know that ψ is onto.

By part(i) and (ii), we know that ψ is isomorphic.

Remark.
The isomorphism is canonical (no choice of basis.) i.e.
φ is natural independes of choice of basis.

Remark.
If dim(V ) = ∞, then V, V ∗ do not size have the same cardinality.

Example.
Let V = {(a1 , a2 , ..., an , ...) : ai ∈ R, an = 0 for all but finitely many } = R⊕N , then we have

(a) V is a vector space over R.

(b) dim(V ) = ∞.

(c) {e1 , e2 , e3 , ...} is a basis for V .

Now, we let V ∗ = {(x1 , x2 , ..., xn , ...) : xi ∈ R} = RN i.e.

X
∞
(x1 , x2 , ..., xn , ...) ←→ l((a1 , a2 , ..., an , ....)) = x i ai
i=1

and we defined the sequence

−
→
α := (α, α2 , α3 , ...) ∈ RN

where α ∈ R, and let S ∈ L(V ∗ , V ∗ ) be defined by

S((x1 , x2 , x3 , ...)) = (x2 , x3 , x4 , ...) i.e. S(−

→
α) = α·−
→
α

Thus we know that the eigenvector − →α of S corresponding to eigenvalue α. Hence, we get {→

−
α }α∈R ,
∗
is linear independent, is a set of elements in V indexed by α ∈ R.
(Since α ∈ R is distinct.)
Conclusion. V ∗ does not have countable basis.
(Ref. 1-1.infinite dimensional vector space by Silberman)

4 | Linear Algebra ReSupply Date X JunWei,Zhang Edit

 Notation. Reference Material
(1) Edition. NTNU Math113 Jun-Wei,Zhang - Date. 2002-06-13.

(2) Linear Algebra (4th Edition 2003) - Stephen H.Friedberg, Arnold J. Insel, Lawrence E. Spence.

(3) Linear Algebra (2nd Edition 1971) - Kenneth M Hoffman, Ray Kunze.

(4) Infinite dimensional vector space by Silberman.

5 | Linear Algebra ReSupply Date X JunWei,Zhang Edit