Week 4
Achim Eduard, Barbu Andreea, Bartes, Cătălin, Bâle Marcu, Capan Matei
March 2020
CHAPTER 2. Differential calculus in Rn
2.1 The normed space of linear mappings
2.1.1 Definition (linear map)
A map ϕ : Rn → Rm is called linear if
∀α, β ∈ R, ∀x, y ∈ Rn : ϕ(αx + βy) = αϕ(x) + βϕ(y)
We make the following notation for the set of all linear maps:
L(Rn , Rm ) := {ϕ : Rn → Rm |ϕ is a linear map}
If ϕ ∈L(Rn , Rm ) then
1◦ ϕ(0n )=0m .
2◦ ϕ(-x)=-ϕ(x).
3◦ ∀k∈ N, ∀ x1 ,..., xn ∈ Rn , ∀α1 ,..., αk ∈ R
ϕ(α1 x1 + ... + αk xk ) = α1 ϕ(x1 ) + ... + αk ϕ(xk )
2.1.2 Theorem (general form of linear maps from Rn to Rm )
A map ϕ : Rn → Rm is linear ⇐⇒ ∃ n vectors v1 ,...,vn ∈ Rm such that
ϕ(x) = x1 v1 + ... + xn vn , ∀x = (x1 , ..., xn ) ∈ Rn
2.1.3 Corollary (general form of linear maps from Rn to R)
A map ϕ : Rn → R is linear ⇐⇒ ∃v∈ Rn such that ϕ(x)=hx,vi, ∀x∈ Rn .
2.1.4 Definition (the matrix of a linear map)
Let ϕ=(ϕ1 ,...,ϕm )∈L(Rn , Rm ). Consider e1 ,...en the vectors from the canonical base in Rn ,
∀i ∈ {1,...n}, ei =(0,...,0,1,0,...,0), where 1 is on the ith position.
Then ϕ(e1 ),...,ϕ(en ) are points of Rm . Let
ϕ(e1 ) := (v11 , v12 , ..., v1m ), v1i = ϕi (e1 ), i = 1, .., m
ϕ(e2 ) := (v21 , v22 , ..., v2m ), v2i = ϕi (e2 ), i = 1, .., m
..
.
ϕ(en ) := (vn1 , vn2 , ..., vnm ), vni = ϕi (en ), i = 1, .., m
We build
υ11 υ21 ··· υn1
υ12 υ22 ··· υn2
[ϕ] := (ϕi (ej ))i=1,m = ..
.. ..
. . .
υ1m υ2m · · · υnm
This is called the matrix of the linear map ϕ.
1
2.1.5 Remark
n
In the
case of matrix equalities, (x1 ,...,xn )∈ R will be identified with the matrix column
x1
..
. ∈ Rn×1 . Thus, we can write
xn
y1
..
hx, yi = (x1 ...xn ) . = xT y
yn
2.1.6 Theorem
If a, b∈ R, and ϕ, ψ ∈L(Rn , Rm ), then aϕ+bψ ∈L(Rn , Rm ) and the equality
[aϕ + bψ] = a[ϕ] + b[ψ]
holds.
2.1.7 Theorem
If ϕ ∈L(Rn , Rm ) and ψ ∈L(Rn , Rm ) then ψ ◦ ϕ ∈L(Rn , Rm ) and the equality
[ψ ◦ ϕ] = [ψ] · [ϕ]
holds.
2.1.8 Theorem
The map ϕ=(ϕ1 ,...,ϕm ):Rn → Rm is linear ⇐⇒ all maps ϕ1 ,...,ϕm :Rn → Rm are linear.
2.1.9 Theorem
Any linear map :Rn → Rm is a Lipschitz map.
2.1.10 Definition (the norm of a linear map)
From theorem 2.1.9 follows that any map ϕ ∈L(Rn , Rm ) is continuous on Rn . Denote
S n−1 := {(x1 , ..., xn )| ∈ Rn |x21 + ... + x2n = 1}
the sphere with the center in the origin and radius 1 from Rn .
Being closed an bounded, S n−1 is a compact subset of Rn .
Because the function ∀x∈ R → ||ϕ(x)|| ∈ [0, ∞) is continuous, according to Weierstrass
theorem we can introduce the real number
||ϕ|| := max
n−1
||ϕ(x)||
x∈S
This is called the norm of the linear map ϕ.
2.1.11 Theorem
If ϕ ∈L(Rn , Rm ) then the next statements are true:
1◦ ||ϕ(x)|| ≤ ||ϕ|| · ||x||, ∀x∈ Rn
2◦ ||ψ ◦ ϕ|| ≤ ||ψ|| · ||ϕ||
2.1.12 Theorem
The linear map || · ||:L(Rn ,Rm )→[0,∞) is a norm of the real linear space L(Rn ,Rm ).
2
2.1.13 Theorem
If ϕ ∈ L(Rn ,Rn ) then, the next propositions are equivalent:
1◦ ϕ is bijective.
2◦ ϕ is injective.
3◦ ϕ is surjective.
4◦ det[ϕ]6=0.
2.1.14 Theorem
If ϕ : Rn → Rn is a bijective linear map, then ϕ−1 ∈L(Rn ,Rm ) and the equality [ϕ−1 ]=[ϕ]−1
holds.
