Mathematics For Business Engineers 1: Homogeneous Functions and Directional Derivatives
Mathematics For Business Engineers 1: Homogeneous Functions and Directional Derivatives
Mathematics For Business Engineers 1: Homogeneous Functions and Directional Derivatives
1/12
Homogeneous functions
Example
Let f : R2 → R : (x, y ) 7→ f (x, y ) = 4x 3 − 5y 3
∀λ ∈ R+ 2
0 , ∀(x, y ) ∈ R : (λx, λy ) ∈ R
2
2/12
Returns to scale
Theorem
Consider a firm with a homogeneous production function q : (R+ )n → R of degree r ∈ R+
0 . Then q
exhibits (a) CRS if r = 1, (b) IRS if r > 1 or (c) DRS if r < 1
Example
The Cobb-Douglas production function q : (R+ )2 → R : (K , L) 7→ q(K , L) = CK α Lβ is
homogeneous of degree α + β
ν
The CES production function q : (R+ )2 → R : (K , L) 7→ q(K , L) = c(αK ρ + (1 − α)Lρ ) ρ is
homogeneous of degree ν
3/12
Euler’s theorem
The proof is a good exercise on applying the chain rule (see book for details)
Example
Consider a homogeneous production function Q : (R+ )2 → R : (K , L) 7→ Q(K , L) of degree r with
continuous partial derivatives
Euler’s theorem ⇒ K ∂Q
∂K
(K , L) + L ∂Q
∂L
(K , L) = rQ(K , L)
Interpretation: the sum of capital times marginal productivity of capital and labour times
marginal productivity of labour is equal to the degree of homogeneity times total production
If Q exhibits CRS (i.e., r = 1), then the result is known as the product exhaustion theorem in
economics
4/12
Directional derivatives
5/12
Geometric interpretation
T
πu
α
p
graph of f
y
u
a
Lu
x
6/12
Example
Example
Let f : R2 → R : (x, y ) 7→ f (x, y ) = x 2 + 3xy , a = (1, 1) and u = (cos θ, sin θ) for some θ ∈ [0, 2π[
Question: Determine Du f (a) using the definition
Solution:
a + hu = (1, 1) + h(cos θ, sin θ) = (1 + h cos θ, 1 + h sin θ)
f (a + hu) = f (1 + h cos θ, 1 + h sin θ) = (1 + h cos θ)2 + 3(1 + h cos θ)(1 + h sin θ)
= h2 (cos2 θ + 3 cos θ sin θ) + h(5 cos θ + 3 sin θ) + 4
f (a) = 1 + 3 = 4
f (a + hu) − f (a)
Du f (a) = lim = lim (h(cos2 θ + 3 cos θ sin θ) + 5 cos θ + 3 sin θ) = 5 cos θ + 3 sin θ
h→0 h h→0
7/12
Alternative formulations of directional derivative
Proof: (a) follows using the appropriate definitions and (b) is a consequence of the chain rule (see
book for details)
8/12
Alternative formulations of directional derivative
Example
Let f : R2 → R : (x, y ) 7→ f (x, y ) = x 2 + 3xy , a = (1, 1) and u = (cos θ, sin θ) for some θ ∈ [0, 2π[
Question: Determine Du f (a) using alternative (a)
Solution:
gu (h) = f (a + hu) = f (1 + h cos θ, 1 + h sin θ) = (1 + h cos θ)2 + 3(1 + h cos θ)(1 + h sin θ)
gu0 (h) = 2(1 + h cos θ) cos θ + 3 cos θ(1 + h sin θ) + 3(1 + h cos θ) sin θ
Du f (a) = gu0 (0) = 2 cos θ + 3 cos θ + 3 sin θ = 5 cos θ + 3 sin θ
Example
Let f : R2 → R : (x, y ) 7→ f (x, y ) = x 2 + 3xy , a = (1, 1) and u = (cos θ, sin θ) for some θ ∈ [0, 2π[
Question: Determine Du f (a) using alternative (b)
Solution:
D1 f (x, y ) = 2x + 3y ⇒ D1 f (1, 1) = 5
D2 f (x, y ) = 3x ⇒ D2 f (1, 1) = 3
Du f (a) = D1 f (a)u1 + D2 f (a)u2 = 5 cos θ + 3 sin θ
9/12
Gradient
Example
Let f : R2 → R : (x, y ) 7→ f (x, y ) = 31 (x cos y + y cos x)
D1 f : R2 → R : (x, y ) 7→ D1 f (x, y ) = 31 (cos y − y sin x)
D2 f : R2 → R : (x, y ) 7→ D2 f (x, y ) = 31 (−x sin y + cos x)
∇f : R2 → R2 : (x, y ) 7→ ∇f (x, y ) = ( 13 (cos y − y sin x), 13 (−x sin y + cos x))
10/12
Visualization of gradient vector field
11/12
Geometric interpretation of the gradient vector
Theorem
Consider f : A ⊂ Rn → R with A an open set of Rn and a ∈ A. Assume that f is partial differentiable
in an open set of A containing a and that all partial derivatives are continuous in a. Then,
(a) the gradient vector ∇f (a) points in the direction where f at a has the largest possible slope
(b) the opposite of the gradient vector −∇f (a) points in the direction where f at a has the
smallest possible slope
(c) a direction perpendicular to the gradient vector ∇f (a) points in the direction where f at a has
no slope
Proof: ∇f (a)
n
X
Du f (a) = Dj f (a)uj = h∇f (a), ui
j=1