Math 208 Exercise Pack
Uploaded by
ChiOfGreeMath 208 Exercise Pack
Uploaded by
ChiOfGreeMath 208 Course Problems Pack
Mathematics for Economics
Exercises By William Thompson
Contents
1 Linear Equations 3
1.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 Algebra of Matrices 13
2.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3 Inverse Matrices and Determinants 26
3.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4 Advanced Topics in Linear Algebra 44
4.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5 Multivariable Regions and Functions 65
5.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
6 Calculus for Functions of n-Variables 71
6.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.3 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.4 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.5 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.6 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7 Optimization of Functions of n-Variables 96
7.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
1
8 Constrained Optimization 104
8.1 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.2 Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
2
Chapter 1
Linear Equations
3
1.1 Practice Questions
1. Consider the following pairs of lines. Graph the lines and solve the system of equations they represent.
Relate the result to the graph.
(a) 2x + y = 4, x − y = 2
(b) −x + y = 1, 3x − 3y = 4
(c) 3x − 5y = 7, 2x + y = 9
(d) 2x − y = 5, 5x − y = 11
2. Solve the system of linear equations
(
x1 − x2 = 0
(a)
3x1 − 2x2 = −1
(
3x1 + x2 = 240
(b)
x1 + 3x2 = 240
9x1 − 3x2 = −1
(c) 1 2 1
x1 + x2 = −
5 5 3
− −
x
1 2 x2 1
+ =2
(d) 4 3
x − 3x = 20
1 2
3. Determine the value(s) of k such that the system of linear equations has the indicated number of solutions.
(
x1 + kx2 = 2
(a) No solution
kx1 + x2 = 4
(
x1 + kx2 = 0
(b) Exactly one solution
kx1 + x2 = 0
(
4x + kx2 = 6
(c) Infinitely many solutions
kx1 + x2 = −3
(
kx1 + x2 = 16
(d) Infinitely many solutions
3x1 − 4x2 = −64
4. Let λ be some real number and consider the system
(
(−2 − λ)x + y = 0
−x − λy = 0
Show that for λ 6= −1 the solution is trivial and that for λ = −1 there are infinite solutions. Find a
parametric representation of this solution.
4
5. Suppose that in an economy the supply and demand of two goods depend on one another and are given by
the equations
( (
q1S = −10 + 3p1 q1D = 40 − p1 + p2
q2S = −7 + 2p1 + p2 q2D = 13 + p1 − 2p2
where qi represents the quantity of goods i = 1, 2. Furthermore pi represents the price of goods i = 1, 2.
Find the quantities of price and supply/demand in equilibrium.
6. There are two markets for goods that are regarded as substitutes and their supply and demand curves are
given by
( (
q1S = 2p1 q1D = 20 − p1 + p2
q2S = −10 + 2p2 q2D = 40 − 2p2 + p1
Find the equilibrium prices and quantities of the two goods.
7. Consider an economy that produces coal and steel. Coal is needed to produce steel and steel is needed to
make the tools to produce coal. Thus the industries are interrelated. Suppose that producing a single unit
(1kg) of coal requires 0.18kg of coal and 0.03kg of steel, while producing a single unit (1kg) of steel requires
0.21kg of coal and 0.04kg of steel.
(a) Let x and y be the total amounts (in kg) or gross output of coal and steel produced and let x∗ and y ∗
be the net amounts (in kg) of coal and steel produced respectively. Find constants a, b, c and d such
that
(
x∗ = ax + by
y ∗ = cx + dy
(b) Solve the system for the gross outputs when the net amounts are x∗ = 26.53 kg of coal and y ∗ = 36.06
kg of steel.
8. An IS-LM graph consists of two curves, IS and LM. Gross domestic product (GDP) is placed on the horizontal
axis, increasing to the right. The interest rate makes up the vertical axis. The IS curve is the curve that slopes
downwards and depicts the set of all levels of interest rates and output (GDP) at which total investment
equals total saving. The LM curve is the curve that slopes upwards and depicts the set of all levels of income
(GDP) and interest rates at which money supply equals money (liquidity) demand. Suppose an economy
has IS and LM curves given by
R = 25 − 2Y
R = −10 + M + Y
2
where R is the interest rate (as a percent), Y is GDP (measured in billions), and M is the money supply. If
the government has a target for GDP of $7.5 billion, what level of money supply will achieve this and what
is the resulting interest rate?
5
Answers
1. (a)
(b)
(c)
(d)
2. (a) x1 = −1, x2 = −1
(b) x1 = 60, x2 = 60
1 2
(c) x1 = − , x2 = −
3 3
(d) x1 = 14, x2 = −2
3. (a) k = 1 and k = −1
6
(b) k 6= ±1
(c) k = −2
3
(d) k = −
4
4. First multiply the second row by (2 + λ) to obtain
(
(−2 − λ)x + y = 0
(−2 − λ)x − λ(2 + λ)y = 0
Then replace the second row by the following (1) − (2) → (2) to obtain
(
(−2 − λ)x + y = 0
(1 + λ(2 + λ))y = 0
By investigating the second equation we see that we may simplify it as
(1 + 2λ + λ2 )y = 0 ⇒ (λ + 1)2 y = 0
by factoring. We see that if λ 6= −1 then we may divide by (1 + λ)2 to obtain y = 0. Then we bounce this
back into the first equation to obtain (−2 − λ)x + 0 = 0. Ultimately since we multiplied the second row by
(−2 − λ) upon elimination we can’t have λ = −2 (since that would mean we multiplied a row by zero). This
means we would obtain x = 0. This is the trivial solution.
On the other hand, say we had λ = −1. Then the system would simplify to
(
−x + y = 0
⇒ −x + y = 0
0y = 0
This is infinite solutions and thus by letting y = t we obtain x = −t, y = t for t ∈ R.
170 30 380 279
5. The prices are p1 = , p2 = . The quantities are q1 = , q2 = .
13 13 13 13
130 170 260 230
6. The prices are p1 = , p2 = . The quantities are q1 = , q2 = .
11 11 11 11
(
x∗ = 0.82x − 0.03y
7. (a)
y ∗ = −0.21x + 0.96y
(b) x = 34 and y = 45.
8. M = 25 and R = 10.
7
1.2 Practice Questions
1. Solve the following systems. State the solutions type as unique, infinite or nonexistent. If it is infinite, find
the parametric form.
x2 + x3 − 2x4 = −3
x + 2x − x = 2
1 2 3
(a)
2x1 + 4x2 + x3 − 3x4 = −2
x − 4x − 7x − x = −19
1 2 3 4
x1 − x2 + 2x3 = 4
x + x = 6
1 3
(b)
2x1 − 3x2 + 5x3 = 4
3x + 2x − x = 1
1 2 3
x − 2y + 3z = 9
(c) −x + 3y = −4
2x − 5y + 5z = 17
(
2x1 + 4x2 − 2x3 = 0
(d)
3x1 + 5x2 = 1
x1 − x2 − x3 = 0
(e) x1 + 2x2 − x3 = 6
2x1 − x3 = 5
3x1 − 2x2 + 4x3 = 1
(f) x1 + x2 − 2x3 = 3
2x1 − 3x2 + 6x3 = 8
2x1 + x2 − 3x3 = 4
(g) 4x1 + 2x3 = 10
−2x1 + 3x2 − 13x3 = −8
(
x1 − 3x2 + 2x3 = 18
(h)
5x1 − 15x2 + 10x3 = 18
2. Which of the following systems are linearly dependent? For those systems that are linearly dependent, find
the nature of dependence. For the rest, find their solutions.
2x + 4y − z = 5
(a) y + z = 2
x+y+z =7
4x − 2y + 2z = 6
(b) −y + z = 1
2x = 2
8
−x + 2y − 3z = 2
(c) −2y = 3
2x − y + z = 9
x+y+z+w =1
−x + y − w = 1
(d)
x−y+z =1
y + z = 1
2x + z − 2w = 5
−x + 2y + z = 4
(e)
y − z + 3w = 1
2x − y + 2z − 5w = 4
3. Determine the value(s) of k such that the system of linear equations has the indicated number of solutions
kx1 + 2kx2 + 3kx3 = 4k
(a) Exactly one solution x1 + x2 + x3 = 0
2x1 − x2 + x3 = 1
x1 + x2 + kx3 = 3
(b) Not a unique solution x1 + kx2 + x3 = 2
kx1 + x2 + x3 = 1
4. Find values of a, b and c (if possible) such that the system of linear equations has (a) a unique solution, (b)
no solution, and (c) infinitely many solutions.
x1 + x2 = 2
x + x = 2
2 3
x 1 + x 3 =2
ax + bx + cx = 0
1 2 3
5. Put the following system of linear equations into a matrix and perform row operations to reduce it to Reduced
Row Echelon Form. Then interpret the result and state the solution.
x1 + 3x2 + 2x3 + 4x4 − 3x5 = −7
2x + 6x − x − 2x = 0
1 2 4 5
6x3 + 2x4 − x5 = 12
x + 3x − x + 4x + 2x = −6
1 2 3 4 5
6. Solve the following system of excess demand functions for p1 , p2 , p3 , and p4 :
10 − p1 − 2p2 + 5p3 − 2p4 = 0
5 − 2p1 + p2 − p3 + 8p4 = 0
6 + 2p1 − p2 − 4p3 − 9p4 = 0
20 − p1 − 2p2 − 2p3 − 2p4 = 0
2
9
7. Find the reduced row echelon form of the following matrices
1 1 1 1
1 1 1 1
(a)
0
1 2 3
0 1 2 3
1 2 1
(b) 2 2 2
1 0 1
1 2 1 2 1
(c) 2 1 2 1 2
0 1 0 1 0
1 2 3
(d)
2 3 4
8. Homogeneous systems naturally occur in many scenarios. One such instance of their emergence is in balancing
chemical equations. For example, suppose we wish to balance the following chemical equation
Fe + Cl2 → FeCl3
Let the underlined regions correspond to the variables x, y and z respectively. Then we denote the concen-
tration of each element as [E]k = k. This gives us the following system of linear equations
( (
x + 0y = z (Equation balancing out iron [Fe]) x−z =0
⇒
0x + 2y = 3z (Equation balancing chlorine [Cl]) 2y − 3z = 0
3
This has the parametric solution x = t, y = t and z = t. Choosing t = 2 gives the smallest positive integer
2
solution, thus balancing the equation and yielding
2 Fe + 3 Cl2 → 2 FeCl3
Use this technique to balance the following equations.
(a) Fe + O2 → Fe2 O3
(b) FeBr3 + H2 SO4 → Fe2 (SO4 )3 + HBr
(c) C4 H6 O3 + H2 O → C2 H4 O2
(d) C 7 H1 6 + O2 → CO2 + H2 O
(e) H2 SiCl2 + H2 O → H8 Si4 O4 + HCl
(f) HSiCl3 + H2 O → H10 Si10 O15 + HCl
(g) C 7 H9 + HNO3 → C7 H6 (NO2 )3 + H2 O
10
Answers
1. (a) x1 = −1, x2 = 2, x3 = 1 and x4 = 3
(b) No solution
(c) x = 1, y = −1 and z = 2
(d) x1 = 2 − 5t, x2 = −1 + 3t and x3 = t for t ∈ R
(e) x1 = 3, x2 = 2, x3 = 1
(f) No solution
5 1
(g) x1 = − t, x2 = 4t − 1 and x3 = t for t ∈ R
2 2
(h) No solution
2. (a) x = 5, y = −3/5 and z = 13/5
(b) Linearly dependent with R1 = 2(R2 + R3 ) on first step of reduction.
(c) x = 11/2, y = −3/2 and z = −7/2
(d) x = 2, y = 1, z = 0 and w = −2
(e) Linearly dependent with R1 = R3 + R4 on first step of reduction.
3. (a) k 6= 0
(b) k = 1, −2
4. (a) a + b + c = 0 (b) a + b + c 6= 0 (c) Not possible
5. The augmented matrix is
1 3 2 4 −3 −7
2
6 0 −1 −2 0
0 0 6 2 −1 12
1 3 −1 4 2 −6
which in RREF is
1 3 0 0 0 1
0 0 1 0 0 3
0 0 0 1 0 −2
0 0 0 0 1 2
which has solution x1 = 1 − 3t, x2 = t, x3 = 3, x4 = −2 and x5 = 2 for t ∈ R.
6. p1 = 8, p2 = 5, p3 = 1 and p4 = 1
1 0 −1 −2
0 1 2 3
7. (a) 0
0 0 0
0 0 0 0
1 0 1
(b) 0 1 0
0 0 0
11
1 0 1 0 1
(c) 0 1 0 1 0
0 0 0 0 0
1 0 −1
(d)
0 1 2
8. (a) 4Fe + 3O2 → 2Fe2 O3
(b) 2FeBr3 + 3H2 SO4 → Fe2 (SO4 )3 + 6HBr
(c) C4 H6 O3 + H2 O → 2C2 H4 O2
(d) C7 H1 6 + 11O2 → 7CO2 + 8H2 O
(e) 4H2 SiCl2 + 4H2 O → H8 Si4 O4 + 8HCl
(f) 10HSiCl3 + 15H2 O → H10 Si10 O15 + 30HCl
(g) C7 H9 + 3HNO3 → C7 H6 (NO2 )3 + 3H2 O
12
Chapter 2
Algebra of Matrices
13
2.1 Practice Questions
1. For the following find x and y.
x −2 −4 −2
(a) =
7 y 7 22
16 4 5 4 16 4 2x + 1 4
(b) −3 13 15 6 = −3 13 15 3x
0 2 4 0 0 2 3y − 5 0
x+2 8 −3 2x + 6 8 −3
(c) 1 2y 2x = 1 18 −8
7 −2 y + 2 7 −2 11
1 2 1 y
(d) =
x−y 2 0 2
3 x 3 y
(e) =
x 1 2 1
3 4 x 3 4 y
(f) =
2 5 7 2 5 7
x 1 3 1
(g) =
y 0 2 0
1 0 0 1 x y
(h) 0 1 1 = 0 x y
0 1 0 x 1 y
2. Solve for x, y and z in the matrix equation
4x 4y 2y + 8 2z + 2x
=
4z −4 −2x + 10 2 − 2x
3. For the following, construct the matrices whose components are given explicitly by the corresponding rule.
(a) The 4 × 4 diagonal matrix D where dii = i2 .
(b) The 4 × 4 matrix A where aij = i + j − 1.
(c) The 3 × 5 matrix B where bij = |i − j|.
(d) The 4 × 4 matrix U where uij = 0 if i > j and uij = i2 − j 2 + 1 if i ≤ j.
(e) The 4 × 4 matrix K where kij = Iij + j − i given that I4 = [Iij ] are the components of the identity
matrix.
(f) The 3 × 3 matrix M where mij = (−1)j 2i .
(g) The 3 × 3 matrix N where nij = (−1)i+j max(i, j).
14
Answers
1. (a) x = −4, y = 22
(b) x = 2 and y = 3
(c) x = −4 and y = 9
(d) x = 2 and y = 2
(e) x = 2 and y = 2
(f) x = t and y = t where t ∈ R (parametric solution)
(g) x = 3 and y = 2
(h) No solution
2. x = 3, y = 2, and z = 1
1 0 0 0
0 4 0 0
3. (a) D = 0 0 9 0
0 0 0 16
1 2 3 4
2 3 4 5
(b) A = 3 4 5 6
4 5 6 7
0 1 2 3 4
(c) B = 1 0 1 2 3
2 1 0 1 2
1 −2 −7 −14
0 1 −4 −11
(d) U =
0 0 1 −6
0 0 0 1
1 1 2 3
−1 1 1 2
(e) K = −2 −1 1 1
−3 −2 −1 1
−2 2 −2
(f) M = −4 4 −4
−8 8 −8
1 −2 3
(g) N = −2 2 −3
3 −3 3
15
2.2 Practice Questions
1
1. For the following find, if possible, (i) A + B, (ii) A − B, (iii) 2A, (iv) 2A − B and (v) B + A
2
1 2 −3 −2
(a) A = and B =
2 1 4 2
2 1 1 2 −3 4
(b) A = and B =
−1 −1 4 −3 1 −2
6 0 3 8 −1
(c) A = and B =
−1 −4 0 4 −3
2. For the following find, if possible, AB and BA
1 2 2 −1
(a) A = and B =
4 2 −1 8
2 −1 3 0 1 2
(b) A = 5 1 −2 and B = −4 1 3
2 2 3 −4 −1 −2
2 1 0 −1 0
(c) A = −3 4 and B = 4 0 2
1 6 8 −1 7
2
(d) A = 3 2 1 and B = 3
0
−1 3
1 2
(e) A = 4 −5 and B =
0 7
0 2
0 −1 0 2
(f) A = 4 0 2 and B = −3
8 −1 7 1
6
−2
(g) A = 1 and B = 10 12
3. Consider the following matrices with indicated size: A3×4 , B3×4 , C4×2 , D4×2 and E4×3 . If defined, determine
the size of the following matrix. If not defined, explain why.
(a) A + B
1
(b) D
2
(c) AC
(d) E − 2A
16
4. Use any method to solve the following system of linear equations. State the system as matrix multiplication
followed by expressing the solution in matrix form when summarizing your answer. Return to these problems
and solve them using the inverse matrix after completing the appropriate sections in Chapter 3.
(
−x1 + x2 = 4
(a)
−2x1 + x2 = 0
(
−2x1 − 3x2 = −4
(b)
6x1 + x2 = −36
x1 − 2x2 + 3x3 = 9
(c) −x1 + 3x2 − x3 = −6
2x1 − 5x2 + 5x3 = 17
x1 − 5x2 + 2x3 = −20
(d) −3x1 + x2 − x3 = 8
−2x2 + 5x3 = −16
2x1 − x2 + x4 = 3
3x − x − x = −3
2 3 4
(e)
x1 + x3 − 3x4 = −4
x + x + 2x = 0
1 2 3
−3 −3 1 3
5. Given that X = and Y = compute X 2 − Y 2 .
5 −6 6 −6
4 0
6. Given that A = compute A3 − 3A2 .
−3 7
7. Which of the following statements is true for all square n × n matrices A and B?
(A) A2 B 2 = A(AB)B
(B) (A − B)2 = A2 − 2AB + B 2
(C) (AB)2 = A2 B 2
(D) (A + B)2 = A2 + 2AB + B 2
(E) (A + B)(A − B) = A2 − B 2
8. For the following show that AB = O (the zero matrix) even though A 6= O and B 6= O
3 3 1 −1
(a) A = and B =
4 4 −1 1
1 −2
" #
2 4
(b) A = and B = 1
2 4 − 1
2
9. This is another example on how the “cancellation law” does not exist in the realm of matrices. For the
following demonstrate that AB = AC even though B 6= C
1 2 3 −8 5 2
A= B= C=
3 6 2 3 1 −2
17
Answers
5
−2 −1
−2 0 4 4 2 4 5 6
1. (a) (i) (ii) (iii) (iv) (v) 5
6 3 −2 −1 4 2 0 0 5
2
5 9
4 −2 5
0 4 −3
4 2 2
2 5 −2
3 −
(b) (i) (ii) (iii) (iv) (v) 7 2 2
−4 0 2 2 −2 6 −2 −2 8 1 −3 10 1
− 0
2 2
12 0 6
(c) (i) Not possible (ii) Not possible (iii) (iv) Not possible (v) Not possible
−2 −8 0
0 15 −2 2
2. (a) AB = , BA =
6 12 31 14
−8 −2 −5 9 5 4
(b) AB = 4 8 17 , BA = 3 11 −5
−20 1 4 −17 −1 −16
3 −4
(c) AB is not possible, BA = 10 16
26 46
6 4 2
(d) AB = [12], BA = 9 6 3
0 0 0
−1 19
(e) AB = 4 −27 , BA is not possible
0 14
3
(f) AB = 10 , BA is not possible
26
60 72
−20 −24
(g) AB = 10
, BA is not possible
12
60 72
3. (a) 3 × 4
(b) 4 × 2
(c) 3 × 2
(d) Undefined as you can only subtract matrices of equal size
−1 −1 x1 4 x1 4
4. (a) = =⇒ =
−2 1 x2 0 x2 8
−2 −3 x1 −4 x1 −7
(b) = =⇒ =
6 1 x2 −36 x2 6
1 −2 3 x1 9 x1 1
(c) −1 3 −1 x2 = −6 =⇒ x2 = −1
2 −5 5 x3 17 x3 2
18
1 −5 2 x1 −20 x1 −1
(d) −3 1 −1 x2 = 8 =⇒ x2 = 3
0 −2 5 x3 −16 x3 −2
1
2 −1 0 1 x1 3 x1 2
1
0 3 −1 −1 x2 −3
=⇒ x2 = − 2
(e) =
1 0 1 −3 x3 −4 x3
0
1 1 2 0 x4 0 x4 3
2
−25 42
5. X 2 − Y 2 =
−15 −33
3 2 16 0
6. A − 3A =
−180 196
7. (A) True (B) False as (A − B)2 = A2 − AB − BA + B 2 (C) False as (AB)2 = (AB)(AB) = ABAB (D) False
as (A + B)2 = A2 + AB + BA + B 2 (E) False as (A + B)(A − B) = A2 + BA − AB − B 2
8. Just compute AB to get the zero matrix.
7 −2
9. AB = AC = even though B 6= C.
21 −6
19
2.3 Practice Questions
1. Compute AT , B T , AT + B T and (A + B)T provided
1 2 −8 4
A= B=
−3 4 −7 5
2. For the following verify that (AB)T = B T AT (where on the left hand side you first compute the product
AB then take its transpose).
−3 0
−1 1 −2
(a) A = and B = 1 2
2 0 1
1 −1
2 1
2 3 1
(b) A = 0 1 and B =
0 4 −1
−2 1
3. For the following find AT A and AAT . Show that each of these products results in a symmetric matrix.
4 2 1
(a) A =
0 2 −1
0 −4 3 2
8 4 0 1
(b) A = −2 3
5 1
0 0 −3 2
4. Let A be a square matrix of size n.
1
(a) Prove that if C = (A + AT ) then C is symmetric.
2
1
(b) Prove that if D = (A − AT ) then DT = −D (called skew symmetric).
2
(c) Prove that any square matrix can be written as the sum of a symmetric and skew symmetric matrix.
(d) Prove that the diagonal of a skew symmetric matrix only has zero entries.
5. The profit function of a firm is given by Π = pT q − wT z where q is the output vector, p is the price per
unit of output vector, z is the input vector and w is the price per unit of input vector. Compute the profit
provided that
2000 10 2000 5
q = 3000 p = 15 z = 2500 w = 10
6000 20 2000 15
6. If possible, compute (Ax)T , xT AT , xxT , xT x and AT xT for the following matrices
1 −3 x1
A= x=
−2 4 x2
20
Answers
T 1 −3 T −8 −7 T T T −7 −10
1. A = ,B = and A + B = (A + B) =
2 4 4 5 6 9
T T T 2 −5
2. (a) (AB) = B A =
4 −1
4 0 −4
(b) (AB)T = B T AT = 10 4 −2
1 −1 −3
16 8 4
T T 21 3
3. (a) A A = 8 8 0 and AA =
3 5
4 0 2
68 26 −10 6 29 −14 5 −5
26 41 3 −1 and AAT = −14 81 −3 2
(b) AT A =
−10 3 43 5 5 −3 39 −13
6 −1 5 10 −5 2 −13 13
T
T 1 T 1 1 1 1
4. (a) C = (A + A ) = (A + AT )T = (AT + (AT )T ) = (AT + A) = (A + AT ) = C
2 2 2 2 2
T
1 1 1 1 1
(b) DT = (A − AT ) = (A − AT )T = (AT − (AT )T ) = (AT − A) = − (A − AT ) = −D
2 2 2 2 2
(c) Let A be any general matrix and let C and D be defined as above.
1 1 1 1 1 1
C + D = (A + AT ) + (A − AT ) = A + AT + A − AT = A
2 2 2 2 2 2
(d) My definition a matrix is skew symmetric if DT = −D. Thus a matrix is skew symmetric if and only
if all its entries satisfy dij = −dji for all i, j = 1, ..., n. On the diagonal i = j and thus we obtain
dii = −dii ⇒ 2dii = 0 ⇒ dii = 0 as desired.
5. Π = 120, 000
x21 x1 x2
6. (Ax)T = x1 − 3x2 −2x1 + 4x2 , xT AT = x1 − 3x2 −2x1 + 4x2 (same), xxT =
,
x2 x1 x22
xT x = x21 + x22 while AT xT is undefined.
21
2.4 Practice Questions
1. Let A be a square matrix that is idempotent. Prove that the matrix I − A is also idempotent.
2 −2 −4
2. Show that the matrix −1 3 4 is idempotent.
1 −2 −3
1/6 −1/3 1/6
3. Show that the matrix −1/3 2/3 −1/3 is idempotent.
1/6 −1/3 1/6
6 −2 −5 1
−2 8 −2 −4 is idempotent.
1
4. Show that the matrix A =
11 −5 −2 6
1
1 −4 1 2
x y
5. Determine all values of x and y such that the matrix is idempotent.
y 1/2
6. If matrices A and B are each partitioned into four submatrices
A11 A12 B11 B12
A= and B=
A21 A22 B21 B22
then you can block multiply A and B provided the sizes of the submatrices are such that the matrix
multiplications and additions are defined.
A11 A12 B11 B12 A11 B11 + A12 B21 A11 B12 + A12 B22
AB = =
A21 A22 B21 B22 A21 B11 + A22 B21 A21 B12 + A22 B22
compute this for the following matrices.
1 2 0
1 2 0 0
and B = −1 1 0
(a) A = 0 1 0 0 0 0 1
0 0 2 1
0 0 3
0 0 1 0 1 2 3 4
0 0 0 1 5 6 7 8
= O2×2 I2
(b) A = and B =
−1 0 0 0 −I2 O2×2 1 2 3 4
0 −1 0 0 5 6 7 8
7. For the matrices A and B below verify that tr(AB) =tr(BA).
3 1
1 −1 1
A= B= 0 1
2 2 0
1 1
8. Find the trace of the following matrices.
22
1 2 3
(a) 0 −2 4
3 1 3
1 0 2 1
0 1 −1 2
(b)
4 2 1 0
0 0 5 1
1 4 3 2
4 0 6 1
(c)
3 6 2 1
2 1 1 −3
9. Prove that each statement is true when A and B are matrices of order n and c is a scalar.
(a) tr(A + B) =tr(A)+tr(B)
(b) tr(cA) = ctr(A)
10. Prove that if AB and BA are square matrices then tr(AB) =tr(BA)
11. Suppose that A and B are matrices such that
1 −2 1
1
AB = −2 4 −2
6
1 −2 1
Use this to compute tr(BABABA).
12. Use the trace to prove that there are no 2 × 2 matrices A and B such that
AB − BA = I
23
Answers
1.
(I − A)2 = I − IA − AI + A2
= I − A − A + A2 Property of Identity matrix
=I −A−A+A Idempotence of A
=I −A+O Property of subtraction
=I −A Property of addition with zero
and thus I − A is idempotent.
2. Confirm this by showing A2 = A.
3. Confirm this by showing A2 = A.
4. Confirm this by showing A2 = A.
1
5. The values must satisfy the equations x2 + y 2 = x, y 2 = 1/4 and xy = y (obtained by setting A2 = A).
2
We reject the possibility that y = 0 from the second equation and thus must have x = 1/2 (from the third
equation) which gives us a corresponding value of y = 1/2 or y = −1/2 from the first and second equation.
This gives us
1/2 1/2 1/2 −1/2
and
1/2 1/2 −1/2 1/2
−1 4 0
6. (a) AB = −1 1 0
0 0 5
1 2 3 4
5 6 7 8
(b) AB =
−1 −2 −3 −4
−5 −6 −7 −8
5 −1 3
4 1
7. tr(AB) =tr = 8 and tr(BA) =tr 2 2 0 = 8.
6 4
3 1 1
8. (a) 2
(b) 4
(c) 0
9. (a)
tr(A + B) = (a11 + b11 ) + (a22 + b22 ) + · · · + (ann + bnn )
= (a11 + a22 + · · · + ann ) + (b11 + b22 + · · · + bnn )
= tr(A) + tr(B)
(b) tr(cA) = ca11 + ca22 + · · · + cann = c(a11 + a22 + · · · + ann ) = ctr(A).
24
10. Let A be of size n × m and B of size m × n. If C = AB then by the definition of matrix multiplication each
m
X
entry of C is given by cij = aik bkj . Thus we obtain that
k=1
tr(AB) = tr(C)
Xn
= cii
i=1
Xn Xm
= aik bki
i=1 k=1
m X
X n
= bki aik
k=1 i=1
= tr(BA)
11. Using the cyclic property tr(BABABA) =tr(ABABAB)=tr((AB 3 )). However AB is idempotent so (AB)3 =
AB and thus tr(BABABA) =tr((AB)3 ) =tr(AB) = 1/6 + 2/3 + 1/6 = 1
12. We derive a contradiction by the following
tr(AB − BA) = tr(I)
⇒ tr(AB) − tr(BA) = 2
using the sum property and the sum of the diagonal for a 2 × 2 identity matrix is 2. Using the cyclic property
tr(AB) =tr(BA) we obtain
⇒ tr(BA) − tr(BA) = 2
⇒0=2
which is a contradiction.
25
Chapter 3
Inverse Matrices and Determinants
26
3.1 Practice Questions
1. For each of the following show that B is the inverse of A (and vice versa)
−1 2 1 −2
(a) if A = and B =
−1 1 1 −1
−2 1
" #
1 2
(b) A = and B = 3 1
3 4 −
2 2
−2 2 3 −4 −5 3
1
(c) A = 1 −1 0 and B = −4 −8 3
3
0 1 4 1 2 0
2 −17 11 1 1 2
(d) A = −1 11 −7 and B = 2
4 −3
0 3 −2 3 6 −5
1 4
2. Let A = . Solve the equation AX = I for a 2 × 2 matrix X.
−1 −3
3. Use the formula for the inverse of a 2 × 2 matrix to compute the inverse matrix of the following.
2 3
(a)
−1 5
−4 −6
(b)
2 3
7 3
−
(c) 21 4
4
5 5
5 0
(d)
0 3
3 2
(e)
−1 1
1 2
(f)
2 4
−k −1 k −2 0 −2
4. Negative exponents of matrices are defined by A = (A ) . Compute A if A =
−1 3
5. In each of the following exercises find a value k such that the matrix is equal to its own inverse
3 k
(a)
−2 −3
2 k
(b)
−1 −2
6. In each of the following exercises find a value k such that the matrix is singular
27
4 k
(a)
−2 −3
k 2
(b)
−3 4
7. Prove that if A is an invertible matrix then (AT )−1 = (A−1 )T .
8. Prove that if A is idempotent then (I − 2A)−1 = I − 2A.
9. Compute the determinant of the following matrices
2 1
(a)
3 4
5 2
(b)
−6 3
−7 6
" #
(c) 1
3
2
0 8
(d)
0 4
3−λ 2
(e)
4 1−λ
6u −1
(f)
−1 3v
2x
e3x
e
(g)
2e2x 3e3x
x ln(x)
(h)
1 1/x
10. Solve the following equations for x
x+3 2
(a)
=0
1 x+2
x−1 2
(b)
=0
3 x−2
11. Let X be a matrix of size n × 2 (hence meaning it is not invertible).
(a) What is the size of X T X and explain why it is possible (but not guaranteed) that this matrix is
invertible.
(b) Suppose that X T X is invertible. Show that P = X(X T X)−1 X T is idempotent.
28
Answers
1. This is true as AB = I and BA = I
2. By computing
1 4 x11 x12 1 0 x11 + 4x21 x12 + 4x22 1 0
= ⇒ =
−1 −3 x21 x22 0 1 −x11 − 3x21 −x12 − 3x22 0 1
which forms a system of linear equations of four variables x11 , x12 , x21 and x22 given by
x11 + 4x21 = 1
x + 4x = 0
12 22
−x 11 − 3x 21 = 0
−x − 3x = 1
12 22
which has the solution x11 = −3, x21 = 1, x12 = −4, and x22 = 1 thus we obtain
−3 −4
X=
1 1
which of course is the inverse of A.
5 3
−
3. (a) 13 13
1 2
13 13
(b) Does not exist
16 15
(c) 594 59
70
−
59 59
1/5 0
(d)
0 1/3
1/5 −2/5
(e)
1/5 3/5
(f) Does not exist
11 3
−2 4 2
4. A = 3 1
4 2
5. (a) k = 4
(b) k = 3
6. (a) k = 6
(b) k = −3/2
29
7. To prove that (A−1 )T = (AT )−1 we want to show the left hand side behaves like the inverse of AT . Multiply
the left hand side with AT to obtain
(A−1 )T AT = (AA−1 )T = (I)T = I
where we have used transpose properties. Showing AT (A−1 )T = I is a similar procedure.
8. We essentially want to show that I − 2A is its own inverse, i.e. (I − 2A)2 = I. We compute to obtain
(I − 2A)2 = I 2 − 2IA − 2AI + 4A2 = I − 2A − 2A + 4A = I
where we have used the fact that IA = AI = A and A2 = A (as it is assumed to be idempotent).
9. (a) 5
(b) 17
(c) −24
(d) 0
(e) λ2 − 4λ − 5
(f) 18uv − 1
(g) e5x
(h) 1 − ln(x)
10. (a) x = −1, −4
(b) x = −1, 4
11. (a) The size of the matrix is n × n. As this is a square matrix it is potentially invertible.
(b)
P 2 = (X(X T X)−1 X T )(X(X T X)−1 X T )
= X(X T X)−1 (X T X)(X T X)−1 X T
= X(X T X)IX T
= X(X T X)X T
=P
and thus P is idempotent.
30
3.2 Practice Questions
1. Compute the determinants of the following matrices
3 0 4
(a) −2 4 1
1 −3 1
1 4 −2
(b) 3 2 0
−1 4 3
2 4 6
(c) 0 3 1
0 0 −5
−0.4 0.4 0.3
(d) 0.2 0.2 0.2
0.3 0.2 0.2
x y −1
(e) 3 2 0
1 1 1
−2 0 0
(f) 4 6 0
−3 7 2
3 0 4
(g) 2 3 2
0 5 −1
2 −4 3
(h) 3 1 2
1 4 −1
2 3 −4
(i) 4 0 5
5 1 6
4 3 0
(j) 6 5 2
9 7 3
λ 2 0
2. Find all values of λ such that 0 λ + 1 2 = 0.
0 1 λ
3. Find the inverse of the following matrices (if they exist)
1 1 1
(a) 3 5 4
3 6 5
31
2 4 6
(b) 2 0 2
6 8 14
1 1 2
(c) 3 1 0
−2 0 3
2 0 0
(d) 0 3 0
0 0 5
0.6 0 −0.3
(e) 0.7 −1 0.2
1 0 −0.9
1 0 0
(f) 3 4 0
2 5 5
2 3 1
(g) 2 −3 −3
4 0 3
4. Use the inverse to solve each system of linear equations
(
x + 2y = −1
(a)
x − 2y = 3
(
x + 2y = 10
(b)
x − 2y = −6
x1 + 2x2 + x3 = 2
(c) x1 + 2x2 − x3 = 4
x2 − 2x2 + x3 = −2
x1 + 2x2 + x3 = 1
(d) x1 + 2x2 − x3 = 3
x2 − 2x2 + x3 = −3
(
5x1 + 4x2 = 2
(e)
−x1 + x2 = −22
−x1 + x2 + 2x3 = 1
(f) 2x1 + 3x2 + x3 = −2
5x1 + 4x2 + 2x3
5 4 x −15
(g) =
−1 1 y −6
0 1 −2 x1 −1
(h) −1 3 1 x2 = 0
2 −2 4 x3 2
32
5. For what values of k does the matrix
2 1 0
1 2 1
0 −3 k
have an inverse?
6. For what values of k are the following matrices singular?
1 0 3
(a) 2 −1 0
4 2 k
0 k 1
(b) k 1 k
1 k 0
33
Answers
1. (a) 29
(b) −58
(c) −30
(d) 0.002
(e) 2x − 3y − 1
(f) −24
(g) 1
(h) −5
(i) −23
(j) 4
2. λ = −2, 0 or 1.
1 1 −1
3. (a) −3 2 −1
3 −3 2
(b) Singular
−3/2 3/2 1
(c) 9/2 −7/2 −3
−1 1 1
1/2 0 0
(d) 0 1/3 0
0 0 1/5
3.75 0 −1.25
(e) 3.4583̄ −1 −1.375
4.16̄ 0 −2.5
1 0 0
(f) −3/4 1/4 0
7/20 −1/4 1/5
3/20 3/20 1/10
(g) 3/10 −1/30 −2/15
−1/5 −1/5 1/5
4. (a) x = 1, y = −1
(b) x = 2, y = 4
(c) x1 = 1, x2 = 1, x3 = −1
(d) x1 = 0, x2 = 1, x3 = −1
(e) x1 = 10, x2 = −12
(f) x1 = 2, x2 = −3, x3 = 3
(g) x = 1, y = −5
(h) x1 = 0, x2 = −1/7, x3 = 3/7
34
5. k 6= −2
6. (a) k = 24
√
(b) k = ± 2/2
35
3.3 Practice Questions
1. Find the determinant of B given |A| = 7 where
a b c a b c
B= d e f A= d e f
3g 3h 3i g h i
2. Find the determinant of B given |A| = 3 where
a b c a b c
B= g h i A= d e f
d e f g h i
3. Find the determinant of B given |A| = 3 where
a b c a b c
B = 3d + a 3e + b 3f + c A= d e f
g h i g h i
4. Use determinant properties to establish the identity
1 x x2
1 y y 2 = (y − x)(z − x)(z − y)
1 z z2
5. Find the determinant of
2 0 1 3 −2
−2 1 3 2 −1
A=
1 0 −1 2 3
3 −1 2 4 −3
1 1 3 2 0
Hint: Look for a row or column with many zeroes and try to create more.
6. Using determinant properties and a combination of methods compute the following determinants
2 3 3 1
0 4 3 −3
(a)
2 −1 −1 −3
0 −4 −3 2
5 1 0 1
1 0 −1 −1
(b)
2 0 1 2
−1 0 3 1
1 −2 7 9
3 −4 5 5
(c)
3 6 1 −1
4 5 3 2
36
1 −1 8 4 2
2 6 0 −4 3
(d) 2 0 2 6 2
0 2 8 0 0
0 1 1 2 3
1 0 1 1
−1 5 −7 4
(e)
−3 5 −9 2
4 4 6 −1
4 −7 9 1
6 2 7 0
(f)
3 6 −3 3
0 7 4 −1
7. Compute the inverses of the following matrices
1 1 1 −1
1 1 −1 1
(a)
1 −1
1 1
−1 1 1 1
1 0 0 0
1 1 0 0
(b)
1 2 1
0
1 3 3 1
1 0 1
8. Compute |A3 | where A = 1 1 2
1 2 1
9. Let A and B be n × n matrices such that AB = I. Prove that |A| =
6 0 and |B| =
6 0
10. Let A and B be n × n matrices such that AB is singular. Prove that either A or B is singular.
11. Use determinant properties to establish the identity
a+b a a
= b2 (3a + b)
a a + b a
a a a+b
12. Let A be a square matrix such that AT A = I. Prove that either |A| = 1 or |A| = −1.
A B
13. Techniques of partitioned matrices also extend to determinants. Let Γ = where
C D
• A and B have the same number of rows;
• C and D have the same number of rows;
• A and C have the same number of columns; and
• B and D have the same number of columns.
37
Provided that A and D are square with |A| =
6 0 then
det(Γ) = det(A) · det(D − CA−1 B)
Use this to appropriately partition and compute the determinants of the following matrices.
1 2 1 0 1 2 1 0
0 1 3 2 0
0 3 2
(a)
2 =
5 0 7 2 5 0 7
2 2 2 2 2 2 2 2
4 0 3 4 0 3
(b) 2 5 0 = 2 5 0
8 2 2 8 2 2
1 2 0 0 1 2 0 0
0 4 0 0 0 4 0 0
(c)
2
=
5 5 7 2 5 5 7
2 2 2 2 2 2 2 2
38
Answers
1. By pulling out a common factor |B| = 3|A| = 21
2. By interchanging rows R2 and R3 , |B| = −|A| = −3
3. Perform the row operation R2 − R1 → R2 to obtain
a b c
3d 3e 3f
g h i
Now take out a common factor from R2 to obtain |B| = 3|A| = 9
4.
1 x x2 ( x2
1 x
1 y y 2 R2 − R1 → R2 y − x2
2
→ 0 y − x
1 z z 2 R3 − R1 → R3
0 z−x z 2 − x2
x2
1 x
Factor Out → (y − x)(z − x) 0 1 y + x
0 1 z+x
x2
1 x
R3 − R2 → R3 → (y − x)(z − x) 0 1 y+x
0 0 z−y
= (y − x)(z − x)(1)(1)(z − y) = (y − x)(z − x)(z − y)
5. Perform the determinant operations R4 + R2 → R4 and R5 − R2 → R5 to obtain the equivalent determinant
2 0 1 3 −2 2 0 1 3 −2
−2 1 3 2 −1 −2 1 3 2 −1
|A| = 1 0 −1 2 3 = 1 0 −1 2 3
3 −1 2 4 −3 1 0 5 6 −4
1 1 3 2 0 3 0 0 0 1
Then expand along the second column to obtain
2 1 3 −2
1 −1 2 3
|A| = +(1)
1 5 6 −4
3 0 0 1
Now apply the column operation C1 −3C4 → C1 (this is only allowed in determinant operations as |A| = |A|T ).
This gives
8 1 3 −2
8
1 3
−8 −1 2 3
−8 −1 2
|A| = = +(1)
13 5 6 −4
13 5 6
0 0 0 1
Now the rest of the work is manageable and |A| = −135
39
6. (a) 8
(b) 8
(c) 136
(d) −1436
(e) 0
(f) −1344
1 1 1 −1
1 1 1 −1 1
7. (a)
4 1 −1 1 1
−1 1 1 1
1 0 0 0
−1 1 0 0
(b)
1 −2 1 0
−1 3 −3 1
8. Since A is square |A3 | = |AAA| = |A| × |A| × |A| = |A|3 = −8
9. Taking the determinant of both sides we obtain |AB| = |I| ⇒ |AB| = 1. By determinant properties
|AB| = |A||B|. Since the product is non-zero then neither determinant is zero.
10. If AB is singular then |AB| = 0. By determinant properties |AB| = |A||B|. Since the product of two
quantities is zero that must mean that one of them is zero.
11.
a+b a a a+b a a
a a + b a = −b b 0
a a a+b −b 0 b
a+b a a
2
= b −1 1 0
−1 0 1
2a + b a a
2
=b 0 1 0
−1 0 1
3a + b a a
= b2
0 1 0
0 0 1
= b2 (3a + b)
where we have used the operations R2 − R1 → R2 , R3 − R1 → R1 , factor out, C1 + C2 → C1 , then
C1 + C3 → C1 .
12. Taking the determinant of both sides we obtain |AT A| = |AT ||A| = |A||A| = |A|2 while on the right hand
side |I| = 1. Thus |A|2 = 1 meaning either |A| = 1 or |A| = −1.
40
13. (a)
−1 !
1 2 0 7 2 5 1 2 1 0
det(Γ) = det det −
0 1 2 2 2 2 0 1 3 2
0 7 2 5 1 −2 1 0
= (1) · det −
2 2 2 2 0 1 3 2
0 7 2 5 −5 −4
= det −
2 2 2 2 3 2
0 7 5 2
= det −
2 2 −4 −4
−5 5
= det
6 6
= −60
(b)
5 0 2 −1
det(Γ) = det([4]) det − [4] 0 3
2 2 8
5 0 2 1
= 4 · det − · 0 3
2 2 8 4
5 0 1 0 6
= 4 · det −
2 2 4 0 24
5 −3/2
= 4 · det
2 −4
= −68
(c)
−1 !
1 2 5 7 2 5 1 2 0 0
det(Γ) = det · det −
0 4 2 2 2 2 0 4 0 0
5 7
= 4 · det
2 2
= −16
41
3.4 Practice Questions
1. Use Cramer’s Rule to solve (if possible) the following systems of linear equations
(
x1 + 2x2 = 5
(a)
−x1 + x2 = 1
(
3x1 + 4x2 = −2
(b)
5x1 + 3x2 = 4
(
20x1 + 8x2 = 11
(c)
12x1 − 24x2 = 21
(
−0.4x1 + 0.8x2 = 1.6
(d)
2x1 − 4x2 = 5.0
4x1 − x2 − x3 = 1
(e) 2x1 + 2x2 + 3x3 = 10
5x1 − 2x2 − 2x3 = −1
3x1 + 4x2 + 4x3 = 11
(f) 4x1 − 4x2 + 6x3 = 11
6x1 − 6x2 = 3
14x1 − 21x2 − 7x3 = −21
(g) −4x1 + 2x2 − 2x3 = 2
56x1 − 21x2 + 7x3 = 7
4x1 − x2 + x3 = −5
(h) 2x1 + 2x2 + 3x3 = 10
5x1 − 2x2 + 6x3 = 1
3x1 − 2x2 + 9x3 + 4x4 = 35
−x − 9x − 6x = −17
1 3 4
(i)
3x3 + x4 = 5
2x + 2x + 8x = −4
1 2 4
2. Use Cramer’s Rule to solve the system of linear equations for x and y
(
kx + (1 − k)y = 1
(1 − k)x + ky = 3
3. For what values of k will the following system have a unique solution? Find the solution of this system using
Cramer’s Rule.
−3x + 5y + 7z = 7
−3x − 7y + kx = 8
15x + 23y − 19z = −40
42
Answers
1. (a) x1 = 1, x2 = 2
(b) x1 = 2, x2 = −2
(c) x1 = 3/4, x2 = −1/2
(d) Does not apply (not a unique solution)
(e) x1 = 1, x2 = 1, x3 = 2
(f) x1 = 1, x2 = 1/2, x3 = 3/2
(g) x1 = 1, x2 = 2, x3 = −1
(h) x1 = −1, x2 = 3, x3 = 2
(i) x1 = 5, x2 = −3, x3 = 2, x4 = −1
4k − 3 4k − 1
2. x = and y = provided k 6= 1/2.
2k − 1 2k − 1
1019 − 361k 57 − 15k 36
3. x = ,y= and z = − and is the only solution provided k 6= 3.
144(k − 3) 144(k − 3) 144(k − 3)
43
Chapter 4
Advanced Topics in Linear Algebra
44
4.1 Practice Questions
1. Find the lengths of the following vectors
1
(a) 2
−1
−1
(b) −2
1
0
(c) 0
−1
1/3
(d) 1/3
1/3
2. Graph the following vectors in standard position
T
(a) u = 2 −4
T
(b) u = −3 −4
T
3. Find the vector v and illustrate the specified vector operations geometrically where u = −2 3 and
T
w = −3 −2 .
3
(a) v = u
2
(b) v = u + 2w
1
(c) v = (3u + w)
2
T T T
4. For the following let u = 1 2 3 ,v= 2 2 −1 , and w = 4 0 −4
(a) Find u − v and v − u.
(b) Find 2u + 4v − w
(c) Find z where 3u − 4z = w
5. Find a unit vector (a) in the direction of u and (b) in the direction opposite that of u
(a) u = (−5, 12)
(b) u = (3, 2, −5)
6. In the following write v as a linear combination of u1 , u2 , and u3 if possible.
T
T T T
(a) v = , u1 = 2 3 5 , u2 = 1 2 4 , u3 = −2 2 3
10 1 4
T T T T
(b) v = 0 5 3 0 , u1 = 1 1 2 2 , u2 = 2 3 5 6 , u3 = −3 1 −4 2
45
7. Argue that the following are not subspaces
(a) {(x, y) : x ≥ 0, y ≥ 0}
(b) The set W of all vectors in R3 whose third component is −1
(c) W is the set of all vectors in R2 whose components are integers.
√
(d) {( t, 0, 3t) : t ≥ 0}
8. Prove that the nullspace of a matrix A is a subspace.
9. Let V and W be two subspaces. Prove that the set {u : u = v + w, v ∈ V and w ∈ W } is a subspace.
10. If possible write each vector as a linear combination of the vectors in S = {(2, −1, 3), (5, 0, 4)}.
(a) z = (−1, −2, 2)
1 27
(b) v = (8, − , )
4 4
(c) w = (1, −8, 12)
(d) u = (1, 1, −2)
11. If possible write each vector as a linear combination of the vectors in S = {(2, 0, 7), (2, 4, 5), (2, −12, 13)}
(a) u = (−1, 5, −6)
(b) v = (−3, 15, 18)
1 4 1
(c) w = ( , , )
3 3 2
(d) z = (2, 20, −3)
12. Determine whether the set S spans R3
(a) S = {(4, 7, 3), (−1, 2, 6), (2, −3, 5)}
(b) S = {(−2, 5, 0), (4, 6, 3)}
(c) S = {(1, −2, 0), (0, 0, 1), (−1, 2, 0)}
13. Determine whether the set S is linearly independent or linearly dependent.
(a) S = {(−2, 2), (3, 5)}
(b) S = {(0, 0), (1, −1)}
(c) S = {(1, −4, 1), (6, 3, 2)}
(d) S = {(−2, 1, 3), (2, 9, −3), (2, 3, −3)}
3 5 3 7 3
(e) S = { , , , 3, 4, , − , 6, 2 }
4 2 2 2 2
(f) S = {(1, 0, 0), (0, 4, 0), (0, 0, −6), (1, 5, −3)}
(g) S = {(0, 0, 0, 1), (0, 0, 1, 1), (0, 1, 1, 1), (1, 1, 1, 1)}
14. Show that each set below is linearly dependent by finding a nontrivial linear combination of vectors in the
set whose sum is the zero vector. The express one of the vectors in the set as a linear combination of the
other vectors in the set (stating the nature of dependence).
46
(a) S = {(3, 4), (−1, 1), (2, 0)}
(b) S = {(1, 1, 1), (1, 1, 0), (0, 1, 1), (0, 0, 1)}
15. For which values of k is each set linearly independent?
(a) S = {(k, 1, 1), (1, k, 1), (1, 1, k)}
(b) S = {(k, 1, 1), (1, 0, 1), (1, 1, 3k)}
16. Let S = {u, v} be a linearly independent set. Prove that {u + v, u − v} is also a linearly independent set.
17. Explain why each set is not a basis for R2 .
(a) S = {(−4, 5), (0, 0)}
(b) S = {(−3, 2)}
(c) S = {(1, 2), (1, 0), (0, 1)}
(d) S = {(6, −5), (12, −10)}
18. Determine a basis of each subspace
(a) W = {(2t, t, −t) : t ∈ R}
(b) W = {(2s − t, s, t, s) : s, t ∈ R}
19. Find the rank of the following matrices
1 0
(a)
0 2
1 −3 2
(b)
4 2 1
1 6 18
(c) 7 40 116
−3 −12 −27
−2 −4 4 5
(d) 3 6 −6 −4
−2 −4 4 9
20. Find the nullspace of each matrix
2 −1
(a)
−6 3
(b) 1 2 3
1 2 3
(c)
0 1 0
1 2 −3
(d) 2 −1 4
4 3 −2
47
5 2
(e) 3 −1
2 1
1 3 −2 4
(f) 0 1 −1 2
−2 −6 4 −8
1 4 2 1
2 −1 1 1
(g)
4 2 1 1
0 4 2 0
21. Consider the following matrices where B is the reduced row echelon form of A
1 1 1 0 0 1 0 3 0 −4
2 5 1 1 0 0 1 −1 0 2
A= B=
3 7 2 2 −2 0 0 0 1 −2
4 9 3 −1 4 0 0 0 0 0
(a) Find the rank and nullity of A
(b) Find a basis for the nullspace of A
(c) Find a basis for the row space of A
(d) Find a basis for the column space of A
(e) Determine whether the rows of A are linearly independent
22. Argue that if a matrix A is not square then either the row vectors of A or the column vectors of A form a
linearly dependent set.
23. Find the dot product of the following pairs of vectors
0 1
(a) 0 and 0
1 0
−1 0
(b) 0 and 0
1 1
1/2 1
(c) 0 and 0
1/2 1
24. Find (u + v) · (2u − v) when u · u = 4, u · v = −5, and v · v = 10.
25. Determine all vectors that are orthogonal to u
(a) u = (0, 5)
(b) u = (2, −1, 1)
48
26. For each of the following show that the set of vectors is pairwise orthogonal and normalize to produce an
orthonormal set.
(a) S = {(−1, 3), (12, 4)}
√ √ √ √ √
(b) S = {( 3, 3, 3), (− 2, 0, 2)}
27. Apply the Gram-Schmidt orthonormalization process to transform the given basis for a subspace of Rn into
an orthonormal basis for the subspace. Use the vectors in the order in which they are given.
(a) S = {(3, 4, 0), (2, 0, 0)}
(b) S = {(4, −3, 0), (1, 2, 0), (0, 0, 4)}
(c) S = {(1, 2, −1, 0), (2, 2, 0, 1), (1, 1, −1, 0)}
49
Answers
√
1. (a) 6
√
(b) 6
(c) 1
√
(d) 3/3
2. (a)
(b)
3. (a)
(b)
50
(c)
T
4. (a) u − v = −1 0 4 = −(v − u)
T
(b) 6 12 6
T
1 3 13
(c) −
4 2 4
5 12 5 12
5. (a) (a) (− , ) (b) ( , − )
13 13 13 13
3 2 5 3 2 5
(b) (a) √ , √ ,− √ (b) − √ ,− √ , √
38 38 38 38 38 38
6. (a) v = u1 + 2u2 − 3u3
(b) It is not possible.
T
7. (a) The vector v = 1 1 is in the space but the scalar multiple −v is not in the space as it contains
negative components.
(b) The sum of two vectors in this space (without any scaling) will have a third component value of −2
and thus not lie in the space.
T
T 1 1 1
(c) The vector v = 1 1 lies in the space but v = is not in the space.
2 2 2
T T
(d) The vector v = 1 0 3 lies in the space but −v = −1 0 −3 does not lie in the space as
no element in the space has a negative first component.
8. A vector x lies in the nullspace of a matrix A if Ax = 0. Let u and v lie in the nullspace, then A(u + v) =
Au + Av = 0 + 0 = 0 and thus the sum lies in the nullspace. Let k be a scalar, then A(ku) = kAu = k0 = 0
and thus any scalar multiple lies in the space.
9. Let x and y lie in the space. Then x = v1 + w1 and y = v2 + w2 for some v1 , v2 ∈ V and w1 , w2 ∈ W .
Then x + y = (v1 + v2 ) + (w1 + w2 ) = v + w where v = v1 + v2 and w = w1 + w2 . As V and W are
subspaces then v ∈ V and w ∈ W and thus x + y lies in the original space. Lastly let k be a scalar, then
kx = (kv1 ) + (kw1 ) and as V and W are subspaces, kv1 ∈ V and kw1 ∈ W and thus kx lies in the original
space.
10. (a) z = 2(2, −1, 3) − (5, 0, 4)
1 3
(b) v = (2, −1, 3) + (5, 0, 4)
4 2
(c) w = 8(2, −1, 3) − 3(5, 0, 4)
(d) u cannot be written as a linear combination of the given vectors.
51
7 5
11. (a) u = − (2, 0, 7) + (2, 4, 5) + 0(2, −12, 13)
4 4
(b) v cannot be written as a linear combination of the given vectors.
1 1
(c) w = − (2, 0, 7) + (2, 4, 5) + 0(2, −12, 13)
6 3
(d) z = −4(2, 0, 7) + 5(2, 4, 5) + 0(2, −12, 13)
12. (a) It spans R3
(b) It does not span R3 , it doesn’t even have enough vectors.
(c) It does not span R3 .
13. (a) Linearly independent
(b) Linearly dependent
(c) Linearly independent
(d) Linearly dependent
(e) Linearly independent
(f) Linearly dependent
(g) Linearly independent
7
14. (a) (3, 4) = 4(−1, 1) + (2, 0)
2
(b) (1, 1, 1) = (1, 1, 0) + (0, 0, 1) − 0(0, 1, 1)
15. (a) All k 6= 1, −2
1
(b) All k 6=
2
16. Set a(u + v) + b(u − v) = 0. Rearrange to obtain (a + b)u + (a − b)v = 0. As S is a linearly independent
set then a + b = 0 and a − b = 0. The solution is unique with a = 0 and b = 0. Thus {u + v, u − v} is a
linearly independent set.
17. (a) It contains the zero vector
(b) There are not enough vectors
(c) There are too many vectors
(d) The set is linearly dependent
T
18. (a) Every vector in the space is of the form v = 2t t −t . We may express this as
2
v = t 1
−1
and thus every vector is given by the span of the set S = {(2, 1, −1)}. As this set is linearly independent
it forms a basis of the subspace.
52
T
(b) Every vector in the space is of the form v = 2s − t s t s . We may express this as
2 −1
1 0
v = s
0 + t 1
1 0
and thus every vector is given by the span of the set S = {(2, 1, 0, 1), (−1, 0, 1, 0)}. You may verify that
this set is a linearly independent set, and thus it forms a basis.
19. (a) Rank 2
(b) Rank 2
(c) Rank 3
(d) Rank 2
1
20. (a) x = t
2
−2 −3
(b) x = t 1 + s 0
0 1
−3
(c) x = t 0
1
−1
(d) x = t 2
1
0
(e) x=
0
2 −1
−2 1
(f) x = t
0 + s
1
1 0
0
0
(g) x= 0
0
21. (a) Rank = 3, Nullity =2
−3 4
1 −2
(b) B = 1 , 0
0 2
0 1
1 0 0
0 1 0
(c) B = 3
,
−1 ,
0
0
0 1
−4 2 −2
53
1 1 0
2 , 5
1
(d) B = ,
3 7 2
4 9 −1
(e) They are linearly dependent. In this case (as there are four rows) we would need Rank(A) = 4 for the
rows to be linearly independent. However Rank(A) = 3 < 4 and so the rows are linearly dependent.
This may also be seen by the fact that there are three vectors required to make a basis (so the dimension
of the rowspace is three). Therefore four rows gives a linearly dependent set.
22. Without loss of generality assume that A is of size m × n with m < n. As the rank (dimension of the row
or column space) is bounded above by m then the dimension of the column space is bounded above by m.
Since you need the dimension of the column space to be n for it to be linearly independent, they must then
be linearly dependent. A similar case holds for the row space where m > n.
23. (a) 0
(b) 1
(c) 1
24. -7
25. (a) All orthogonal vectors are of the form (t, 0).
(b) All orthogonal vectors are of the form (t, s, −2t + s)
1 1
26. (a) √ (−1, 3), √ (3, 1)
10 10
1 1 1 1 1
(b) √ ,√ ,√ , − √ , 0, √
3 3 3 2 2
3 4 4 3
27. (a) , ,0 , ,− ,0
5 5 5 5
4 3 3 4
(b) ,− ,0 , , , 0 , (0, 0, 1)
5 5 5 5
( √ √ √ ! √ √ √ ! √ √ √ !)
6 6 6 3 3 3 3 3 3
(c) , ,− ,0 , , 0, , , ,− ,− ,0
6 6 6 3 3 3 3 3 3
54
4.2 Practice Questions
1. For the following matrices (a) find the characteristic equation and (b) the eigenvalues with corresponding
eigenvectors.
6 −3
(a)
−1 1
1 2
(b)
2 1
3
1 −
(c) 1
2
−1
2
2 −2 3
(d) 0 3 −2
0 −1 2
1 2 −2
(e) −2 5 −2
−6 6 −3
0 −3 5
(f) −4 4 −10
0 0 4
2 0 0 0
0 2 0 0
(g)
0 0 3 1
0 0 4 0
2. Find the eigenvalues of the following matrices
−4 5
(a)
−2 3
1 1
(b) 21 3
1
− −
3 3
2 4 2
(c) 1 0 1
1 −4 5
1
3 − 5
2
1 1 1
(d)
− − −
3 6 4
0 0 4
3. Suppose that we have the matrix P = X(X T X)−1 X T
(a) Show that P is idempotent (note that X might not be square! Thus you can’t use the inverse properties
for products)
55
(b) Prove that all possible eigenvalues of an idempotent matrix are either λ = 1 or λ = 0. Hint: Start off
with the eigenvalue problem Av = λv and consider what happens when you multiply both sides on the
left by A.
(c) Determine the algebraic multiplicity of λ = 1 and λ = 0 of P provided
1 2
1 4
X=
1
1
1 3
using the fact that the trace is given by the sum of eigenvalues. Hint: Use the cyclic property of the
trace. Be sure to keep track of the size of matrices.
a b
4. Consider the matrix . If you are given that the eigenvalues are λ = 0, 1 what are the possible values
0 d
of a, b and d?
5. Diagonalize the following matrices (if possible)
6 −3
(a)
−2 1
2 −2 3
(b) 0 3 −2
0 −1 2
1 2 −2
(c) −2 5 −2
−6 6 −3
1 0 0
(d) 1 2 1
1 0 2
6. Orthogonally diagonalize the following symmetric matrices. If you encounter an eigenvalue that has multiple
eigenvectors you will need to perform the Gram-Schmidt procedure to make them orthonormal.
−2 2
(a)
2 1
0 10 10
(b) 10 5 0
10 0 −5
2 2 −2
(c) 2 −1 4
−2 4 −1
7. Use diagonalization to compute the following powers
10 18
(a) A = . Compute A6
−6 −11
56
2 0 −2
(b) A = 0 2 −2 . Compute A5
3 0 −3
8. Prove that the product of orthogonal matrices (of the same size) is an orthogonal matrix.
57
Answers
1. (a) (a) λ(λ − 7) = 0 (b) λ = 0, (1, 2); λ = 7, (3, −1)
(b) (a) (λ + 1)(λ − 3) = 0 (b) λ = −1, (−1, 1); λ = 3, (1, 1)
1 1 1
(c) (a) λ2 − = 0 (b) λ = − , (1, 1); λ = , (3, 1)
4 2 2
(d) (a) (λ − 2)(λ − 4)(λ − 1) = 0 (b) λ = 4, (7, −4, 2); λ = 2, (1, 0, 0); λ = 1, (−1, 1, 1)
(e) (a) (λ + 3)(λ − 3)2 = 0 (b) λ = −3, (1, 1, 3); λ = 3, (1, 0, −1), (1, 1, 0)
(f) (a) λ − 4)(λ − 6)(λ + 2) = 0 (b) λ = −2, (3, 2, 0); λ = 4, (5, −10, −2); λ = 6, (1, −2, 0)
(g) (a) (λ − 2)2 (λ − 4)(λ + 1) = 0 (b) λ = 2, (1, 0, 0, 0), (0, 1, 0, 0); λ = 4, (0, 0, 1, 1); λ = −1, (0, 0, 1, −4)
2. (a) λ = −2, 1
1 1
(b) λ = − ,
6 3
(c) λ = −1, 4, 4
√
17 ± 385
(d) λ = 4,
12
3. (a)
P 2 = (X(X T X)−1 X T )(X(X T X)−1 X T )
= X(X T X)−1 (X T X)(X T X)−1 X T
= XI(X T X)−1 X T
= X(X T X)−1 X T
=P
(b) Multiplying both sides of Av = λv we obtain A2 v = A(λv) = λAv. Using the fact that A2 = A and
Av = λv we obtain
Av = λ(λv)
Using the fact that Av = λv once again we obtain
λv = λ2 v
and as v =6= 0 we must have λ2 = λ and thus λ = 0 or λ = 1.
(c) We compute tr(P ) =tr(X(X T X)−1 X T )=tr(X T X(X T X)−1 ) =tr(I2 ) = 2. Since P is of size 4 × 4 and
λ1 + λ2 + λ3 + λ4 = 2 where either λi = 0 or λi = 1 then we have that the algebraic multiplicity of
λ = 1 is 2 and the algebraic multiplicity of λ = 0 is also 2.
4. The determinant and trace are given by a + d = 1 + 0 = 1 and ad − 0b = 1 × 0 = 0. Thus a + d = 1 and
ad = 0. This is independent of b and thus b may be any value. From the above we have either a = 0, d = 1
or a = 1, d = 0.
6 −3 1 −3 0 0 1/7 3/7
5. (a) =
−2 1 2 1 0 7 −2/7 1/7
58
2 −2 3 −1 1 7 1 0 0 0 1/3 2/3
(b) 0 3 −2 = 1 0 −4 0 2 0 1 3/2 −1/2
0 −1 2 1 0 2 0 0 4 0 −1/6 1/6
1 2 −2 1 1 −1 −3 0 0 1/3 −1/3 1/3
(c) −2 5 −2 = 1 1 0 0 3 0 −1/3 4/3 −1/3
−6 6 −3 3 0 1 0 0 3 −1 1 0
(d) This is not diagonalizable as it has an eigenvalues with geometric multiplicity less than its algebraic
multiplicity.
√ √ √ √
−2 2 −2/√ 5 1/√5 −3 0 −2/√ 5 1/√5
6. (a) =
2 1 1/ 5 2/ 5 0 2 1/ 5 2/ 5
0 10 10 −2/3 1/3 2/3 −15 0 0 −2/3 1/3 2/3
(b) 10 5 0 = 1/3 −2/3 2/3 0 0 0 1/3 −2/3 2/3
10 0 −5 2/3 2/3 1/3 0 0 15 2/3 2/3 1/3
√ √
−2/3
2 2 −2 1/3 2/√5 −2/3√ 5 −6 0 0 1/3
√ √ 2/3
(c) 2 −1 4 = −2/3 1/ 5 4/3√5 0 3 0 2/ √ 5 1/ √5 0√
−2 4 −1 2/3 0 5/3 5 0 0 3 −2/3 5 4/3 5 5/3 5
6 −188 −378
7. (a) A =
126 253
2 0 −2
(b) A5 = −30 32 −2
3 0 −3
8. It is enough to show it for two orthogonal matrices. Let Q1 and Q2 be orthogonal matrices of the same size.
We wish to show that the inverse of Q1 Q2 is (Q1 Q2 )T . We compute the following using transpose properties
and the fact that Q1 QT1 = I and Q2 QT2 = I,
(Q1 Q2 )(Q1 Q2 )T = (Q1 Q2 )(QT2 QT1 )
= Q1 (Q2 QT2 )QT1
= Q1 IQT1
= Q1 QT1
=I
You may compute (Q1 Q2 )T (Q1 Q2 ) = I as well using a similar procedure. Thus the product of two orthogonal
matrices is orthogonal. Now if Q = Q1 Q2 · · · Qn represent the product of any number of orthogonal matrices
and if P is an orthogonal matrix you may prove P Q is orthogonal by generalizing the above procedure.
59
4.3 Practice Questions
1. Determine the definiteness of the matrix associated to the following quadratic forms
(a) g(x1 , x2 ) = x21 + 2x1 x2
(b) g(x − 1, x2 ) = −x21 + 4x1 x2 − 4x22
(c) g(x1 , x2 ) = −x21 + 2xy − 3x22
(d) g(x1 , x2 ) = 4x21 + 8x1 x2 + 5x22
(e) g(x1 , x2 ) = −x21 + x1 x2 − 3x22
(f) g(x1 , x2 ) = x21 − 6x1 x2 + 9x22
(g) g(x1 , x2 ) = 4x21 − x22
1 1
(h) g(x1 , x2 ) = x21 − x1 x2 + x22
2 4
(i) g(x1 , x2 ) = 6x1 x2 − 9x22 − x21
2. Determine the definiteness of the matrix associated to the following quadratic forms
(a) g(x1 , x2 , x3 ) = −x21 − x22 − 2x23 + 2x1 x2
(b) g(x1 , x2 , x3 ) = x21 − 2x1 x2 + x1 x3 + 2x2 x3 + 2x23 + 3x1 x3
(c) g(x1 , x2 , x3 ) = −4x21 − x22 + 4x1 x3 − 2x23 + 2x2 x3
(d) g(x1 , x2 , x3 ) = −x21 − x22 + 2x1 x3 + 4x2 x3 + 2x23
(e) g(x1 , x2 , x3 ) = −x21 + 2x1 x2 − 2x22 + 2x1 x3 − 5x23 + 2x2 x3
(f) g(x1 , x2 , x3 ) = x22 + x1 x2 + 2x1 x3
(g) g(x1 , x2 , x3 ) = −3x21 + 2x1 x2 − x22 + 4x2 x3 − 8x23
(h) g(x1 , x2 , x3 ) = 2x21 + 2x1 x2 + 2x22 + 4x23
3. Consider the quadratic form g(x1 , x2 , x3 ) = 2x21 + 2x1 x3 + 2kx2 x3 + 2x23 . Determine the definiteness of the
matrix associated to this quadratic form for each possible value of k.
4. Determine the values of k for which the quadratic form g(x1 , x2 , x3 ) = x21 + 2kx1 x2 + 2x1 x3 + x23 is positive
definite, negative definite, positive semidefinite, negative semidefinite, and indefinite.
5. Consider the matrix below
s 1 t
1 −1 0
t 0 −2
Find conditions on s and t under which the matrix is negative definite, negative semidefinite, positive definite,
positive semidefinite, and indefinite.
6. Argue that the following matrix
1 0 1 1
0 1 0 0
1 0 1 0
1 0 0 1
is not positive definite by computing only one minor.
60
7. Determine the values of k such that the quadratic form g(x1 , x2 , x3 ) = x21 − 2kx1 x2 − 2x1 x3 + x22 + 4x2 x3 + 5x23
is positive definite. At what values of k is it positive semi-definite but not positive definite?
8. Let g(x1 , x2 , x3 ) = kx21 + 4kx22 + 4kx23 + 4x1 x2 + 2kx1 x3 + 4x2 x3 .
(a) Determine the values of k for which it is positive definite.
(b) Determine the values of k for which it is negative definite.
(c) Determine the values of k for which it is positive semi-definite but not positive definite.
(d) Determine the values of k for which it is negative semi-definite but not negative definite.
(e) Determine the values of k for which it is indefinite.
9. Let g(x1 , x2 , x3 ) = 9x21 + 3x22 + x23 + 2kx1 x3
(a) Determine the values of k for which it is positive definite.
(b) Determine the values of k for which it is positive semi-definite but not positive definite.
(c) Is it possible for the quadratic form to be negative definite or negative semi-definite?
(d) Determine the values of k for which it is indefinite.
10. Determine conditions on a and b for which the quadratic form g(x1 , x2 , x3 ) = x21 + 4x22 + bx23 + 2ax1 x2 + 2x2 x3
is positive definite. Note: you want to construct an inequality that relates a and b. There are no explicit
numerical solutions.
11. If you wish to use the eigenvalues of a matrix to determine the definiteness you might not be able to solve
for the eigenvalues by hand. Instead you must use a numerical calculator to approximate the eigenvalues.
Geogebra’s CAS calculator has a way of approximating the eigenvalues of a matrix. Open up the CAS
calculator, then the command
eigenvalues({{0.2,0.1,-0.3},{0.1,0.2,-3.2},{0.3,-3.2,0.5}})
0.2 0.1 −0.3
approximates the eigenvalues of the symmetric matrix 0.1 0.2 −3.2 . Use this command to approx-
0.3 −3.2 0.5
imate the eigenvalues and determine the definiteness of the following matrices.
0.2 0.1 −0.3
(a) 0.1 0.2 −3.2
0.3 −3.2 0.5
3.01 0.01 1.14
(b) 0.01 1.23 0.13
1.14 0.13 2.54
2.31 −1.12 −1.07
(c) −1.12 2.22 1.31
−1.07 1.31 2.71
−3.16 2.01 0.54
(d) 2.01 −2.37 −0.17
0.54 −0.17 −2.89
61
1.31 −1.26 2.15 −0.11
−1.26 4.19 −1.73 1.91
(e)
2.15 −1.73 6.22 −2.36
−0.11 1.91 −2.36 4.20
62
Answers
1. (a) Indefinite
(b) Negative semidefinite
(c) Negative definite
(d) Positive definite
(e) Negative definite
(f) Positive semidefinite
(g) Indefinite
(h) Indefinite
(i) Negative semidefinite
2. (a) Negative semidefinite
(b) Indefinite
(c) Negative semidefinite
(d) Indefinite
(e) Negative semidefinite
(f) Indefinite
(g) Negative definite
(h) Positive definite
3. If k = 0 the matrix is positive semidefinite. If k 6= 0 the matrix is indefinite.
4. If k 6= 0 the matrix is indefinite. If k = 0 the matrix is positive semidefinite. No other definiteness states
can be obtained.
5. There are no values of s and t for which the matrix is positive definite or positive semidefinite. It is negative
definite if s < −1 and 2s + 2 + t2 < 0. It is negative semidefinite if s ≤ −1 and 2s + 2 + t2 ≤ 0. Otherwise
the matrix is indefinite.
6. The second order principal minor obtained by deleting the second and fourth rows and columns is the
determinant of the zero matrix. Thus the matrix is not positive definite as all such determinant minors need
to be positive.
7. It is positive definite for 0 < k < 4/5. It is positive semi-definite but not positive definite when k = 0 or
k = 4/5.
8. (a) k > 1
(b) k < −1
(c) k = 1 or k = 0
(d) k = −1 or k = 0
(e) −1 < k < 0 or 0 < k < 1
9. (a) −3 < k < 3
(b) k = 3 or k = −3
63
(c) It is not possible as the first term satisfies |A1 | = 9 > 0
(d) k > 3 or k < −3
10. It is positive definite if 4 − a2 > 0 and 4b − 1 − a2 b > 0. Thus simpler conditions are given by −2 < a < 2
1
and b > .
4 − a2
11. (a) λ1 ≈ −2.8, λ2 ≈ 0.2 and λ3 ≈ 3.5. The matrix is indefinite as some of the eigenvalues differ in sign.
(b) λ1 ≈ 1.21, λ2 ≈ 1.64 and λ3 ≈ 3.94. The matrix is positive definite as all the eigenvalues are positive.
(c) λ1 ≈ 1.05, λ2 ≈ 1.42 and λ3 ≈ 4.77. The matrix is positive definite as all the eigenvalues are positive.
(d) λ1 ≈ −4.95, λ2 ≈ −2.78 and λ3 ≈ −0.70. The matrix is negative definite as all the eigenvalues are
negative.
(e) λ1 ≈ 0.11, λ2 ≈ 2.67, λ3 ≈ 3.49, and λ4 ≈ 9.65. The matrix is positive definite as all the eigenvalues
are positive.
64
Chapter 5
Multivariable Regions and Functions
65
5.1 Practice Questions
1. The consumption set of a consumer is C = {(x, y) ∈ R2+ : x ≥ x0 , y ≥ y 0 }. Illustrate this set. Is it closed?
Bounded? Convex?
2. A consumer’s budget set is B = {(x, y) ∈ R2+ : p1 x + p2 y ≤ m} where p1 , p2 > 0 are prices and m > 0 is
income. Illustrate this set. Is it closed? Bounded? Convex?
3. Graph the domain of the function f (x1 , x2 ) = ln(1 − x21 − x22 ). Determine if the set is open, closed or neither.
Determine if the set is bounded or unbounded. Determine if the domain is non-strict convex, strict convex
or neither. Can you guarantee this function has an absolute maximum or minimum without graphing it?
q
4. Graph the domain of the function f (x1 , x2 ) = x2 − x21 . Determine if the set is open, closed or neither.
Determine if the set is bounded or unbounded. Determine if the domain is non-strict convex, strict convex
or neither. Can you guarantee this function has an absolute maximum or minimum without graphing it?
p
5. Graph the domain of the function f (x1 , x2 ) = 1 − |x1 | − |x2 |. Determine if the set is open, closed or
neither. Determine if the set is bounded or unbounded. Determine if the domain is non-strict convex,
strict convex or neither. Can you guarantee this function has an absolute maximum or minimum without
graphing it?
1
6. Graph the domain of the function f (x1 , x2 ) = Determine if the set is open, closed or neither.
(x21
+ − 1)3/2x22
Determine if the set is bounded or unbounded. Determine if the domain is non-strict convex, strict convex
or neither. Can you guarantee this function has an absolute maximum or minimum without graphing it?
7. Find the Euclidean distance between the following points
(a) (−6, 2) and (8, −1)
(b) (5, −3, 0, 8) and (12, −6, 3, 1)
8. Find convex combinations of the following pairs of points fully simplified.
(a) (−1, 1) and (3, 4)
(b) (−2, 0, 1) and (1, −2, 2)
66
Answers
1. It forms a rectangle (that contains its edges) in the first quadrant with an the right corner infinitely far away.
Thus it is closed, unbounded and convex.
2. The set is closed, bounded and convex as it forms a triangle in the first quadrant. You can’t guarantee the
function has an absolute maximum or minimum because the domain is not closed. It may or may not.
3. The domain is given by the open disc x21 + x22 < 1. The set is open. The set is bounded. The set is strictly
convex. You can’t guarantee the function has an absolute maximum or minimum because the domain is not
closed. It may or may not.
4. The domain is given by the upper part of the closed parabola x2 ≥ x21 . The set is closed. The set is
unbounded. The set is strictly convex. You can’t guarantee the function has an absolute maximum or
minimum because the domain is not bounded. It may or may not.
5. The domain is given by the closed shape |x1 | + |x2 | ≤ 1. It is possible to graph this determining all regions
where either x1 + x2 ≤ 1, −x1 + x2 ≤ 1, x1 − x2 ≤ 1 and −x1 − x2 ≤ 1. The resulting shape is a a filled in
diamond (with edges included) that has vertices (1, 0), (0, 1), (−1, 0), and (0, −1). The set is closed. The set
is bounded. The set is non-strict convex. You are able to guarantee the function obtains both an absolute
maximum and minimum because the function is continuous and the domain is both closed and bounded.
6. The domain is the open shape x21 + x22 > 1. The set is open. The set is unbounded. The set is not convex.
You can’t guarantee the function has an absolute maximum or minimum because the domain is neither
closed nor open. It may or may not.
√
7. (a) 205
√
(b) 116
8. (a) (3 − 4t, 4 − 3t)
(b) (1 − 3t, −2 + 2t, 2 − t)
67
5.2 Practice Questions
1. Sketch typical level sets of the following functions and state whether they are (strictly) quasiconcave or
(strictly) quasiconvex. Then say whether the functions are concave, convex, or neither.
(a) f (x1 , x2 ) = 2x21 − x1 x2 + 2x22 Hint: You might need graphing software for this one.
r
1 2 1 2
(b) f (x1 , x2 ) = x + x
2 1 2 2
1/2 1/2
(c) f (x1 , x2 ) = 2x1 x2
2. Sketch typical level sets, then their better and worse sets, of the following functions and state whether they
are quasiconcave.
(a) f (x1 , x2 ) = x1 x2
(b) f (x1 , x2 ) = x21 + x22
(c) f (x1 , x2 ) = −x21 − x22
3. Construct an example of a strictly quasiconcave function that is not a concave function.
√
4. Using the definition, prove that f (x) = x is a concave function.
p √ √
That is, argue that tx0 + (1 − t)x00 > t x0 + (1 − t) x00
p √ √
Hint: It’s easier to show this by proving ( tx0 + (1 − t)x00 )2 − (t x0 + (1 − t) x00 )2 > 0.
5. By drawing a picture argue that the sum of two quasiconcave function is not necessarily quasiconcave.
6. Argue (using the definition of convexity and concavity) that a function defined on a convex set S is concave
if and only if the set of points on or below its graph is a convex set, i.e. {(x, y) : x ∈ S and y ≤ f (x)} is
convex. Similarly the function is convex if and only if the set of points on or above its graph is a convex set,
i.e. {(x, y) : x ∈ S and y ≥ f (x)} is convex.
7. Either by using a theorem or the definition prove that the following sets are convex. If a set is not convex
provide a counterexample.
(a) {(x, y) : y = ex }
(b) {(x, y) : y ≥ ex }
(c) (x, y) : xy ≥ 1, x > 0, y > 0}
Hint: Consider two points xy ≥ 1 and uv ≥ 1 and show the convex combination satisfies [x, u][y, u] ≥ 1.
Upon expansion you may rewrite uy + xv = uv(v/y) + xy(v/y), then use xy ≥ 1 and uv ≥ 1. Then
form the inequality 0 ≤ (y − v)2 ⇒ y 2 + v 2 ≥ 2yv and relate it to the previous inequality.
68
Answers
1. (a) Strictly quasiconvex; convex
(b) Strictly quasiconvex; convex
(c) Strictly quasiconcave; concave
2. (a) The better set is not convex (reminder there is no restriction on both x and y being positive) and thus
is not quasiconcave. As an example, consider the level set associated to (x1 , x2 ) = (1, 1). The points
(1, 1) and (−1, 1) lie in the set but the point (0, 0) which lies on the convex combination of these two
points is not in the set.
(b) Graphically the worse set of this function is the entire plane with a circle centered at the origin being
cut out. This is not a convex set and thus the function is not quasiconcave.
(c) This is a quasiconcave function as the better set of this function at any point is a circle that is filled in.
3. There are several counterexamples. One such single variable function is f (x) = 1 − x2/3 which you may see
by graphing. A multivariable example is f (x1 , x2 ) = x21 x22
4. We find the following
√ √ √ √
( tx0 + (1 − t)x00 )2 − (t x0 + (1 − t) x00 )2 = tx0 + (1 − t)x00 − (t2 x0 + 2t(1 − t) x0 x00 + (1 − t)2 x00 )
p
√ √
= t(1 − t)x0 + t(1 − t)x00 − 2t(1 − t) x0 x00
√ √
= t(1 − t)(x0 − 2 x0 x00 + x00 )
√ √
= t(1 − t)( x0 − x00 )2 ≥ 0
p √ √ 2 p √
Therefore
√ 2 ( tx +p
0 (1 − t)x00 )2 − (t x0 √ ) > 0 which gives ( tx0 + (1 − t)x00 )2 > (t x0 + (1 −
+ (1 − t) x00√
t) x00 ) or rather tx0 + (1 − t)x00 > t x0 + (1 − t) x00 as desired.
5.
6. Here’s a formal argument but I’m happy if you argue by a picture since I don’t expect students in this class
to have formal experience in proof writing. We’ll just prove the first one statement as the second one is
similar. Let L = {(x, y) : x ∈ S and y ≤ f (x)}. First suppose f is concave and let (x, y) ∈ L and (x0 , y 0 ) ∈ L.
Our goal is to prove the convex combination lies in L. This in turn implies that x, x0 ∈ S and y ≤ f (x),
y 0 ≤ f (x0 ). These two inequalities imply that
69
(1 − t)y + ty 0 ≤ (1 − t)f (x) + tf (x0 )
for any 0 ≤ t ≤ 1. Now, because S is convex we obtain that (1−t)x+tx0 ∈ S, so that this convex combination
is in the domain of f . Now because f is concave we obtain
(1 − t)f (x) + tf (x0 ) ≤ f ((1 − t)x + tx0 )
and thus
(1 − t)y + ty 0 ≤ f ((1 − t)x + tx0 )
and hence ((1 − t)x + tx0 , (1 − t)y + ty 0 ) = (1 − t)(x, y) + t(x0 , y 0 ) ∈ L establishing that L is a convex set.
Going the other direction, assume that L is a convex set. Let x, x0 ∈ S. Then (x, f (x)) ∈ L and (x0 , f (x0 )) ∈
L, so by the convexity of L,
(1 − t)(x, f (x)) + t(x0 , f (x0 )) = ((1 − t)x + tx0 , (1 − t)f (x) + tf (x0 ))
lies in L for any 0 ≤ t ≤ 1. Thus (1 − t)f (x) + tf (x0 ) ≤ f ((1 − t)x + tx0 ) establishing that f is concave.
7. (a) This is not a convex set as etu+(1−t)v 6= teu + (1 − t)ev . As a counter example consider the points in the
set (0, 1) and (1, e). Then we may see that for t = 1/2 we have
1 1 1 1
(1) + e 6= e 2 (0)+ 2 (1)
2 2
(b) Using the solution to the first exercise, as f (x) = ex is a convex function then the region lying above
its graph is a convex set.
(c) Let xy ≥ 1 and uv ≥ 1. We want to show that
(tx + (1 − t)u)(ty + (1 − tv)) ≥ 1
By expanding we obtain
(tx + (1 − t)u)(ty + (1 − tv)) = t2 xy + t(1 − t)(uy + xv) + (1 − t)2 uv
Now as uy + xv = uv(y/v) + xy(v/y) ≥ y/v + v/y (using the fact that xy ≥ 1 and uv ≥ 1). Also notice
that by expanding the inequality 0 ≤ (y −v)2 = y 2 −2yv +v 2 we see that 2yv ≤ y 2 +v 2 ⇒ 2 ≤ y/v +v/y.
Thus uy + xv ≥ 2 and thus
t2 xy + t(1 − t)(uy + xv) + (1 − t)2 uv ≥ t2 + 2t(1 − t) + (1 − t)2 = (t + (1 − t))2 = 1
70
Chapter 6
Calculus for Functions of n-Variables
71
6.1 Practice Questions
x1
1. Using the limit definition of the derivative, compute the partial derivatives of f (x1 , x2 ) = .
1 + x1 x2
2. Demonstrate that for the function f (x1 , x2 ) = |x1 |x2 the partial derivative f1 (0, 1) does not exist while
f2 (0, 1) exists.
3. Compute the partial derivatives f1 , f2 of the following functions
(a) f (x1 , x2 ) = 2x31 + x1 x2
x1 + 2
(b) f (x1 , x2 ) =
x2 + 1
1/3 2/3
(c) f (x1 , x2 ) = 9x1 x2
(d) f (x1 , x2 ) = 2x21 − 3x2 − 4
q
(e) f (x1 , x2 ) = x21 + x22
x1 + x2
(f) f (x1 , x2 ) =
x1 x2 − 1
(g) f (x1 , x2 ) = ln(1 + ex1 x2 )
4. Compute the partial derivatives f1 , f2 and f3 of the following functions
(a) f (x1 , x2 , x3 ) = 1 + x1 x22 − 2x23
q
(b) f (x1 , x2 , x3 ) = x1 − x22 + x23
(c) f (x1 , x2 , x3 ) = e−x1 x2 x3
(d) f (x1 , x2 , x3 ) = x2 x3 ln(x1 x2 )
5. Demonstrate that the law of diminishing marginal productivity holds for the Constant of Elasticity (CES)
function y = A(δx−r −r −1/r
1 + (1 − δ)x2 ) where A > 0, 0 < δ < 1 and r > −1. Hint: Use the fact that
r+1 r+1 −r −r −(1/r)−1
y = A (δx1 + (1 − δ)x2 ) .
6. Using the multivariable chain rule, evaluate the following functions. Let g(u) = f (x(u))
(a) Let f (x1 , x2 , x3 ) = x1 x2 + x2 x3 + x1 x3 where x1 = u1 + u2 , x2 = u1 − u2 and x3 = u1 u2 . Compute gu1
and gu2 at the point (u1 , u2 ) = (1/2, 1)
x1 − x2
(b) Let f (x1 , x2 , x3 ) = where x1 = u1 + u2 + u3 , x2 = u1 − u2 + u3 , x3 = u1 + u2 − u3 . Compute
x2 − x3 √
gu1 , gu2 , and gu3 at the point (u1 , u2 , u3 ) = ( 3, 2, 1)
(c) Let f (x1 , x2 ) = x21 + (x2 /x1 ) where x1 = u1 − 2u2 + 1, x2 = 2u1 + u2 − 2. Compute gu2 at the point
(u1 , u2 ) = 0, 0)
dz
7. Find the derivative given
dt
(a) z = exy + 2xy 2 + t where x = 3t, y = 1 − t
(b) z = ln(u) + vt where u = 2t2 and v = t + 1
72
8. For the following state the general multivariable chain rule formula with the given function composition
dz
(a) where z = f (x, y), x = g(t), and y = h(t)
dt
∂w
(b) where w = h(x, y, z), x = f (u, v), y = g(u, v), z = k(u, v)
∂u
∂w
(c) where w = g(x, y), x = h(u, v), y = k(u, v)
∂v
∂y
(d) where y = f (u), u = g(r, s)
∂r
∂w
(e) where w = f (x, y), x = g(r), y = h(s)
∂s
dw
(f) where w = f (x, y, t), x = t2 , y = ln(t)
dt
9. Use implicit differentiation to find the value of dy/dx at the given point.
(a) x3 − 2y 2 + xy = 0 at (1, 1)
(b) x2 + xy + y 2 − 7 = 0 at (1, 2)
(c) (x3 − y 4 )6 + ln(x2 + y 2 ) = 1 at (−1, 0)
10. Use implicit differentiation to find zx and zy at the point (1, 1, 1) provided z 3 − xy + yz + y 3 − 2 = 0
73
Answers
1. By the limit definition
f (x1 + h, x2 ) − f (x1 , x2 )
f1 (x1 , x2 ) = lim
h→0 h
x1 +h x1
1+(x1 +h)x2 − 1+x1 x2
= lim
h→0
h
1 (x1 + h)(1 + x1 x2 ) − [1 + (x1 + h)x2 ]x1
= lim
h→0 h [1 + (x1 + h)x2 ](1 + x1 x2 )
1 x1 + x1 x2 + h + hx1 x2 − x1 − x21 x2 − hx1 x2
2
= lim ·
h→0 h [1 + (x1 + h)x2 ](1 + x1 x2 )
1 h
= lim ·
h→0 h [1 + (x1 + h)x2 ](1 + x1 x2 )
1
= lim
h→0 [1 + (x1 + h)x2 ](1 + x1 x2 )
1 1
= =
[1 + (x1 + 0)x2 ](1 + x1 x2 ) (1 + x1 x2 )2
and the other partial derivative is given by
f (x1 , x2 + h) − f (x1 , x2 )
f2 (x1 , x2 ) = lim
h→0 h
x1 x1
1+x1 (x2 +h) − 1+x1 x2
= lim
h→0
h
1 x1 (1 + x1 x2 ) − [1 + x1 (x2 + h)]x1
= lim ·
h→0 h [1 + x1 (x2 + h)](1 + x1 x2 )
1 x1 + x21 x2 − x1 − x21 x2 − hx21
= lim ·
h→0 h [1 + x1 (x2 + h)](1 + x1 x2 )
x21
= lim −
h→0 [1 + x1 (x2 + h)](1 + x1 x2 )
x21 x21
=− =−
[1 + x1 (x2 + 0)](1 + x1 x2 ) (1 + x1 x2 )2
2. It is simple enough to compute f2 (x1 , x2 ). We treat x1 as constant to obtain
∂ ∂
f2 (x1 , x2 ) = [|x1 |x2 ] = |x1 | [x2 ] = |x1 | · 1 = |x1 |
∂x2 ∂x2
and then we have f2 (0, 1) = |0| = 0 and so f2 (0, 1) exists. Now for the other partial derivative. However, we
need to use the limit definition to show a function is not differentiable and so we set this up as
74
f (0 + h, 1) − f (0, 1)
f1 (0, 1) = lim
h→0 h
f (h, 1) − f (0, 1)
= lim
h→0 h
|h|(1) − |0|(1)
= lim
h→0 h
|h|
= lim
h→0 h
This is a standard function in Calculus I. This is the step function g(t) = |t|/t which simplifies as g(t) = 1
if t > 0 and g(t) = −1 if t < 0. Therefore there is a jump discontinuity where h = 0 and so the above limit
does not exist. Since the limit does not exist then f1 (0, 1) does not exist.
3. (a) f1 = 6x21 + x2 , f2 = x1
1 x1 + 2
(b) f1 = , f2 = −
x2 + 1 (x2 + 1)2
−2/3 2/3 1/3 −1/3
(c) f1 = 3x1 x2 , f2 = 6x1 x2
(d) f1 = 4x1 , f2 = −3
x1 x2
(e) f1 = p 2 2
, f2 = p 2
x1 + x2 x1 + x22
1 + x22 1 + x21
(f) f1 = − , f2 = −
(x1 x2 − 1)2 (x1 x2 − 1)2
x x
x2 e 1 2 x 1 e 1 x2
x
(g) f1 = , f2 =
1 + ex1 x2 1 + e x1 x2
4. (a) f1 = x22 , f2 = 2x1 x2 , f3 = −4x3
x2 x
(b) f1 = 1, f2 = − p 2 , f3 = − p 3
x2 + x23 x22 + x23
(c) f1 = −x2 x3 e−x1 x2 x3 , f2 = −x1 x3 e−x1 x2 x3 , f3 = −x1 x2 e−x1 x2 x3
x2 x3
(d) f1 = , f2 = x3 ln(x1 x2 ) + x3 , f3 = x2 ln(x1 x2 )
x1
∂y y r+1 δ ∂y y r+1 1 − δ
5. Using the hint, the marginals are given by = · and = · r+1 . We see that as xi
∂x1 Ar xr+11
∂x2 Ar x2
∂y
increases then decreases in either case.
∂xi
6. (a) gu1 (1/2, 1) = 3, gu2 (1/2, 1) = −3/2
√ √ √
(b) gu1 ( 3, 2, 1) = 0, gu2 ( 3, 2, 1) = 1, gu3 ( 3, 2, 1) = −2
(c) gu2 (0, 0) = −7
dz
7. (a) = e3t(1−t) (3 − 6t) + 6(1 − t)2 − 12t(1 − t) + 1
dt
dz 2
(b) = + 2t + 1
dt t
75
dz
8. (a) = fx (x, y)g 0 (t) + fy (x, y)h0 (t)
dt
∂w
(b) = hx (x, y, z)fu (u, v) + hy (x, y, z)gu (u, v) + hz (x, y, z)ku (u, v)
∂u
∂w
(c) = gx (x, y)hv (u, v) + gy (x, y)kv (u, v)
∂v
∂y df ∂g
(d) = f 0 (u)gr (r, s) =
∂r du ∂r
∂w ∂f dh
(e) = fx (x, y)(0) + fy (x, y)h0 (s) = fy (x, y)h0 (s) =
∂s ∂y ds
dw 1 fy (x, y, t)
(f) = fx (x, y, t)(2t) + fy (x, y, t) · + ft (x, y, t)(1) = 2tfx (x, y, z) + + ft (x, y, t)
dt t t
9. (a) dy/dx = 4/3
(b) dy/dx = −4/5
(c) dy/dx = 20
10. zx = 1/4, zy = −3/4
76
6.2 Practice Questions
1. Compute the gradient and Hessian of the following functions
(a) f (x1 , x2 ) = x1 + x2 + x1 x2
(b) f (x1 , x2 ) = ln(4x1 + 3x2 )
(c) f (x1 , x2 ) = x1 ex2 + x2 + 1
2
(d) f (x1 , x2 ) = x2 ex1 −x2
x1 − x2
(e) f (x1 , x2 ) = 2
x1 + x2
1
(f) f (x1 , x2 , x3 ) = ex3 + + x1 e−x2
x1
(g) f (x1 , x2 , x3 ) = x21 e3x2 +x1 +x3 + 2x32 /x1
(h) f (x1 , x2 , x3 ) = Axa1 xb2 xc3 (Cobb-Douglas)
1
2. Consider the function f (x) = c + aT x + xT Cx where a and x are size n × 1 and C is size n × n. Show that
2
the gradient and Hessian are given by
∇f (x) = a + Cx ∇2 f (x) = C
3. Which order of differentiation will calculate fxy faster: x first or y first? Try to answer without writing
anything down.
x
(a) f (x, y) = y +
y
(b) f (x, y) = y + x2 y + 4y 3 − ln(y 2 + 1)
(c) f (x, y) = x ln(xy)
u2 − v 2
4. Consider w = f (x, y) where x = and y = uv. Compute an expression for wuu + wvv in terms of
2
sums of fxx , fxy and fyy using the procedure of the second order chain rule.
5. Let g(t) = f (x1 (t), x2 (t)) where f (x1 , x2 ) is a twice continuously differentiable function and x1 (t) = at,
x2 (t) = bt. Compute d2 g/dt2 in terms of a, b and the second partial derivatives of f .
77
Answers
1 + x2 2 0 1
1. (a) ∇f = ,∇ f=
1 + x1 1 0
4
16 12
− −
(4x1 + 3x2 )2 (4x1 + 3x2 )2
(b) ∇f = 4x1 + 3x2 2
, ∇ f =
3 12 9
− −
4x1 + 3x2 (4x1 + 3x2 )2 (4x1 + 3x2 )2
ex2 ex2
0
(c) ∇f = x2 , ∇2 f = x2
x1 e + 1 e x 1 e x2
" 2
# " 2 2
#
2x1 x2 ex1 −x2 2 2(1 + 2x21 )x2 ex1 −x2 2x1 (1 − x2 )ex1 −x2
(d) ∇f = 2 ,∇ f= 2 2
(1 − x2 )ex1 −x2 2x1 (1 − x2 )ex1 −x2 (x2 − 2)ex1 −x2
x2 − x21 + 2x1 x2
(x21 + x2 )2
(e) ∇f = 2
,
x1 + x1
− 2 2
(x1 + x2 )
(−2x1 + 2x2 )(x21 + x2 ) − 2x1 (x2 − x21 + 2x1 x2 ) 2x31 + 3x21 − 2x1 x2 − x2
(x21 + x2 )3 (x21 + x2 )3
∇2 f =
3 2
2x1 + 3x1 − 2x1 x2 − x2 2(x21 + x1 )
(x21 + x2 )3 (x21 + x2 )3
1 2
−x2
− x2 + e 2 x3 −e−x2 0
(f) ∇f = 1 ,∇ f = 1
−x1 e−x2 −e−x2 x1 e−x2 0
ex3 0 0 ex3
(g)
(2x1 + x21 )e3x2 +x1 +x3 − 2x32 /x21
∇f = 3x21 e3x2 +x1 +x3 + 6x22 /x1
2 3x2 +x1 +x3
x1 e
(2 + 4x1 + x21 )e3x2 +x1 +x3 + 4x32 /x31 3(2x1 + x21 )e3x2 +x1 +x3 − 6x22 /x21 (2x1 + x21 )e3x2 +x1 +x3
∇2 f = 3(2x1 + x21 )e3x2 +x1 +x3 − 6x22 /x21 9x21 e3x2 +x1 +x3 + 12x2 /x1 3x21 e3x2 +x1 +x3
2 3x1 +x2 +x3 2 3x2 +x2 +x3 3x2 +x1 +x3
(2x1 + x1 )e 3x1 e e
Aax1a−1 xb2 xc3 Aa(a − 1)xa−2 b c
Aabxa−1 b−1 c
Aacxa−1 b c−1
1 x2 x3 1 x2 x3 1 x2 x3
(h) ∇f = Abxa1 xb−1
2 x3
c
, ∇2 f = Aabxa−1 b−1 c
1 x2 x3 Ab(b − 1)xa1 xb−2
2 x3
c
Abcxa1 xb−1
2 x3
c−1
a b c−1 a−1 b c−1 a b−1 c−1 a b c−2
Acx1 x2 x3 Aacx1 x2 x3 Abcx1 x2 x3 Ac(c − 1)x1 x2 x3
2. This is a fairly standard exercise in computing all the first and second partials of f then arranging them into
a gradient and Hessian. It might be helpful to use specific vectors in R2 by
a1 x1 c12 c12
a= x= C=
a2 x2 c21 c22
and expanding everything to get an idea of how to compute it in general.
3. (a) x first
78
(b) x first
(c) y first
4. wu = fx xu + fy yu = ufx + vfy and similarly wv = fx xv + fy yv = −vfx + ufy . Then applying the product
rule and multivariable chain rule on fx and fy when differentiating wu with respect to u and wv with respect
to v we obtain
wuu = fx + u2 fxx + v 2 fyy = wvv = −fx + v 2 fxx + u2 fyy
⇒ wuu + wvv = (u2 + v 2 )(fxx + fyy )
5. dg/dt = af1 (at, bt) + bf2 (at, bt) ⇒ d2 g/dt2 = a2 f11 (at, bt) + 2abf12 (at, bt) + b2 f22 (at, bt)
79
6.3 Practice Questions
3/2
1. Consider the constant elasticity demand function Q = 2p−2 1 p2 where Q is the demand and pi is the price
of good i for i = 1, 2. Suppose current prices are p1 = 6 and p2 = 9.
(a) What is the current demand for Q?
(b) Use the total differentials to estimate the change in demand p1 increase by 0.20 and p2 decreases by 0.5
(c) Estimate the total demands after the change in (b) and compare your estimates with the actual demand
2. Consider a production function given by p(x1 , x2 ) = 2x21 x22 + 4x1 with current level of input x1 = 10, x2 = 5.
What is the current production? If we increase input 1 by 4 units, estimate how many units input 2 will
change to keep the same level of production.
3. Find the total differential of the function f (x1 , x2 ) = x1 x2 and show that the isoquants are strictly convex
to the origin. Here the domain of the function is assumed to be R2+ .
4. Find the total differential for the function f (x, y) = x1/2 y. Compute and compare the total differential and
the actual change of the function value when (x, y) changes from (1, 1) to (4, 2) and from (1, 1) to (1, 68, 1.20).
5. Argue that for each production function below the isoquants in R2+ are convex to the origin
(a) p(x1 , x2 ) = x2 (1 + x1 )
(b) p(x1 , x2 ) = x1 x2 ex1
6. Find the MRS of the utility function
y = A(δx−r −r −1/r
1 + (1 − δ)x2 )
7. Construct a production function whose isoquants are perfectly substitutable (meaning increasing the input
of one variable by k amount always decreases the input of the other by m amount regardless of where you
are on the curve).
8. Here are some exercises that test your creativity in using implicit differentiation, chain rule, or the differential.
(a) The demand and supply for a good both depend upon the price p of the good and the tax rate t. They
are given by D = f (p, t) and S = g(p, t). For any given value of t, and equilibrium price is a solution
of the equation f (p, t) = g(p, t). Assume that this equation defines p as a differentiable function of t.
dp
Find in terms of the partial derivatives of f and g.
dt
(b) The demand for a good both depends upon the price p of the good and the tax rate t. The supply of
the good depends on the price. Specifically, D = f (p, h(t)) and S = g(p). For any given value of t, an
equilibrium price is a solution of the equation f (p, h(t)) = g(p). Assume that this equation defines p as
dp
a differentiable function of p. Find in terms of the derivatives of f , g, and h.
dt
80
Answers
1. (a) The current demand if Q(6, 9) = 1.5
(b) Using Q1 (6, 9) = 0.5, Q2 (6, 9) = 0.25 we find that ∆Q ≈ (−0.5)(0.20) + (0.25)(−0.5) = −0.225 and
thus the demand is reduced by approximately 0.225
(c) The total demand after the change in (b) is estimated to be 1.5 - 0.225 = 1.275. Since the new prices
are p1 = 6 + 0.20 = 6.20 and p2 = 9 − 0.5 = 8.5 the actual demand after the change is
G(6.2, 8.5) ≈ 1.289
2. The current production is p(10, 5) = 2(100)(25) + 4(10) = 5040 and thus has an associated isoquant of
2x21 x22 + 4x1 = 5040. Use implicit differentiation to obtain
dx2 4x1 x22 + 4
=−
dx1 4x21 x2
At this current level the slope gives dx2 /dx1 |(10,5) = −1004/2000. Therefore with ∆x1 = 4 we obtain
1004
∆x2 ≈ − ∆x1 = −2.008. Thus we decrease input 2 by approximately 2 units.
2000
3. The total differential is
df = x2 dx1 + x1 dx2
Thus the derivative along any level curve is given by
dx2 x2
=− <0
dx1 x1
By taking the second derivative (with respect to x1 ) and using dx2 /dx1 = −x2 /x1 we obtain
dx2
d2 x2 x1 dx1
− x2 (1) x1 (−x2 /x1 ) − x2 2x2
2 =− 2 =− 2 = 2 >0
dx1 x1 x1 x1
and thus the isoquants are strictly convex to the origin.
An alternate way to solve this is to fix an output C to form a generic isoquant C = x1 x2 . This value of C is
positive as all values of x1 and x2 are positive. Now solve for x2 to obtain that the curve in the x1 x2 -plane
(first quadrant) is given by
C
x2 =
x1
and now the first derivative gives us
dx2 C
=− 2 <0
dx1 x1
and the second derivative gives
81
d2 x2 2C
2 = 3 >0
dx1 x1
therefore the isoquant is convex to the origin.
4. The first order total differential is
1
df = fx (x, y)dx + fy (x, y)dy = x−1/2 ydx + x1/2 dy
2
From (1, 1) to (4, 2) the approximate and real change is df = 5/2 = 2.5 and ∆f = f (4, 2) − f (1, 1) = 3. From
(1, 1) to (1.68, 1.20) the approximate change is df = 0.54 and the real change is ∆f = f (1.68, 1.20)−f (1, 1) ≈
0.555.
5. (a) For any output C the isoquants are given by x2 (1+x2 ) = C. We rearrange this to obtain x2 = C/(1+x1 ).
We then may compute
dx2 C
=− <0
dx1 (1 + x1 )2
d2 x2 2C
= >0
dx21 (1 + x1 )3
meaning they are convex to the origin.
(b) For any output C the isoquants are given by x2 x1 ex1 = C. We rearrange this to obtain x2 = Ce−x1 /x1 .
We then may compute
dx2 e−x1 (1 + x1 )
= −C <0
dx1 x21
d2 x2 x21 (−e−x1 (1 + x1 ) + e−x1 ) − 2x1 e−x1 (1 + x1 ) e−x1 (x21 + x1 + 2)
= −C = C >0
dx21 x41 x31
meaning they are convex to the origin.
−r−1
δ x1
6. MRS =
1 − δ x2
7. Any linear function of the form p(x1 , x2 ) = ax1 + bx1 . The differential of this curve is 0 = ∆p ≈ a∆x1 +
b∆x2 ⇒ −b∆x2 ≈ a∆x1 and each 1/a unit increase in x1 leads to a 1/b unit decrease in x2 . However, since
the function is linear the approximation is in fact, exact, and not an approximation. You may demonstrate
this behaviour by graphing the isoquants.
8. (a) Differentiating both sides of the equilibrium equation we obtain
dp dp
fp + ft = gp + gt
dt dt
and solve for the derivative gives
dP gt (p, t) − ft (p, t)
=
dt fp (p, t) − gp (p, t)
82
Note that one could get the same result using differentials
fp dp + dt dt = gp dp + gt dt
and solving for dp/dt.
(b) Using the same technique as before
dp fh (p, h(t))h0 (t)
= 0
dt g (p) − fp (p, h(t))
83
6.4 Practice Questions
1. All the results in this section only help establish a function is convex or concave if it is twice differentiable. If
your function is not differentiable, then you will need to go back to the definition of convexity and concavity
to determine its shape. Argue that the following functions are not differentiable and use the definition of
convexity and concavity to determine whether they are convex or concave.
(a) The function f (x) = |x1 | + |x2 | + · · · + |xn | on Rn . Hint: Use the triangle inequality |a + b| ≤ |a| + |b|.
(b) The Leontief utility function f (x) = min(x1 , x2 , ..., xn ) on Rn+ . Hint: min(x1 , x2 , ..., xn ) = xi for some
xi .
2. Suppose that we have a non-differentiable function and want to establish that it is neither convex nor concave.
You could use the definition of convexity and concavity to show that the inequalities are not satisfied for some
given points and values. An alternate plan is to demonstrate that the function is neither quasiconvex nor
quasiconcave using the better and worse sets. Since every convex function is quasiconvex, then disproving
quasiconvexity also disproves convexity! Argue that the following non-differentiable functions are neither
convex nor concave by disproving that they are quasiconvex and quasiconcave.
2/3 2/3
(a) f (x1 , x2 ) = x1 + x2 on R2 (you will need to use GeoGebra to graph the level curves).
(b) f (x1 , x2 ) = |x1 + x2 | − |x1 | on R2 (you will need to use GeoGebra to graph the level curves).
1/4 1/3
3. Determine whether the profit function π = 4x1 x2 − x1 − x2 is convex, concave, strictly convex, strictly
concave or none of the above on the domain R2+
4. Use the Hessian to determine the curvature of the function
f (x, y) = 2(x − y)2 + x4 + y 4
5. Use the Hessian to determine the curvature of the function
f (x1 , x2 ) = e3x1 +x2
6. Use the definiteness of the Hessian matrix to determine if the following functions are convex, concave or
neither on R2+
(a) f (x1 , x2 ) = x21 x22
1/2 1/2
(b) f (x1 , x2 ) = x1 x2
1/4 1/2
(c) f (x1 , x2 ) = x1 x2
−1/2 −1/2
(d) f (x1 , x2 ) = x1 x2
−2 −2
(e) f (x1 , x2 ) = x1 x2
7. Assume that f and g are both twice differentiable concave functions. Argue that h(x) = af (x) + bg(x) is a
concave function provided a, b ≥ 0.
8. For each of the following functions, determine which, if any, of the following conditions the function satisfies:
Concavity, strict concavity, convexity, strict convexity. (Use whatever technique is most appropriate for each
case)
84
(a) f (x, y) = x + y
(b) f (x, y) = x2
(c) f (x, y) = x + y − ex − ex+y
(d) f (x, y) = x2 − xy + y 2 + 3x − 2y + 1
(e) f (x, y, z) = x2 + y 2 + 3z 2 − xy + 2xz + yz
1/3 1/2
9. Suppose that a firm that uses 2 inputs has the production function f (x1 , x2 ) = 12x1 x2 and faces the
inputs prices (p1 , p2 ) and the output price 1. Show that f is concave over R2+ . Then show that the firm’s
profit function π(x1 , x2 ) = f (x1 , x2 ) − p1 x1 − p2 x2 has the same Hessian, implying it is concave as well.
1
10. Let f (x1 , x2 ) = x31 + 2x21 + 2x1 x2 + x22 − 8x1 − 2x2 − 8 Find the domain of values of (x1 , x2 ) for which the
2
function is convex, if any.
11. Determine the values of k (if any) for which the function f (x, y, z) = 2x2 + 2xz + 2kyz + 2z 2 is concave and
for which it is convex.
12. Show that the function f (x1 , x2 , x3 , x4 ) = −x24 + 2x4 x1 − x21 − x22 + 4x2 xz − x23 is not concave.
13. The following functions are not necessarily convex or concave unless we restrict their domain so that their
Hessian is definite. Determine the domains of the plane where the following functions are convex or concave.
(a) f (x1 , x2 ) = (x1 − 1)2 + x1 x22
1
(b) f (x1 , x2 ) = x31 − 4x1 x2 + 12x1 + x22
3
(c) f (x1 , x2 ) = e−x1 + e−x2
(d) f (x1 , x2 ) = ex1 x2
√
(e) f (x1 , x2 ) = ln( x1 x2 )
85
Answers
1. (a) First, this function is non-differentiable at the origin for any partial derivative (it is also not differentiable
at other places in the domain, but this is the easiest place to demonstrate). To demonstrate this consider
computing fi (0). By definition of the derivative we have
f (0, ..., 0 + h, ..., 0) − f (0, ..., 0, ..., 0)
fi (0) = lim
h→0 h
|0| + · · · + |0 + h| + · · · + |0| − (|0| + · · · + |0| + · · · + |0|)
= lim
h→0 h
|h|
= lim
h→0 h
this is a famous step function encountered in Calculus 1. The limit does not exist due to a jump
|h| |h|
discontinuity given by the below and above limits lim = −1 and lim = 1. Now that we have
h→0 − h h→0 + h
established that f (x) is not differentiable, we aim to show it is convex using the definition. We first note
that f (x) is an additive function, and thus can show it is convex if each term individually is convex.
Consider the single variable function g(x) = |x|. By the triangle inequality
g(tx + (1 − t)y) = |tx + (1 − t)y|
≤ |tx| + |(1 − t)y|
= |t| · |x| + |1 − t| · |y|
= t|x| + (1 − t)|y|
= tg(x) + (1 − t)g(y)
where we have used the fact that 0 ≤ t ≤ 1 so that t is positive. Thus by definition g(x) is convex.
Now, the function of interest is an additive function of the form
f (x) = g(x1 ) + g(x2 ) + · · · + g(xn )
Since each of the functions is convex then the additive function f (x) is convex.
(b) The function has a corner whenever all the components are equal (and other spots too, but this is
the easiest to demonstrate). To demonstrate this consider computing fi (1, ..., 1). By definition of the
derivative we have
min(1, ..., 1 + h, ..., 1) − min(1, ..., 1, ..., 1)
fi (1, ..., 1) = lim
h→0 h
min(1, ..., 1 + h, ..., 1) − 1
= lim
h→0 h
Now, if h < 0 then the below limit is given by
min(1, ..., 1 + h, ..., 1) − 1 1+h−1 h
lim = lim = lim =1
h→0− h h→0 − h h→0 h
−
while if h > 0 then the above limit is given by
86
min(1, ..., 1 + h, ..., 1) − 1 1+h−1 1−1
lim = lim = lim = lim 0 = 0
h→0+ h h→0− h h→0+ h h→0+
and so the limit does not exist as there is a jump discontinuity. Now that we have established that
f (x) is not differentiable, we aim to show it is concave using the definition. Using the hint we have the
following
f (tx + (1 − t)y) = min(tx1 + (1 − t)y1 , ..., txn + (1 − t)yn )
= txi + (1 − t)yi For some i
≥ t min(x1 , ..., xn ) + (1 − t) min(y1 , ..., yn )
= tf (x) + (1 − t)f (y)
where we have used the fact that t is positive as 0 < t < 1. Thus by definition f (x) is concave.
2. (a) The graph below gives the level curve associated to an output of C = 1 with the associated better set
in green and the worse set in blue.
We can see that neither the blue nor green region forms a convex region in R2 . Take a line segment near
the tips of the star to disprove the worse set is convex and take a line segment between two points on
opposite sides of the star to disprove the better set is convex. Since neither the worse set nor better set is
a convex region then the function is neither quasiconvex nor quasiconcave. Since every convex function
2/3 2/3
is quasiconvex and every concave function is quasiconcave then the function f (x1 , x2 ) = x1 + x2 is
neither convex nor concave.
(b) The graph below gives the level curve associated to an output of C = 1 with the associated better set
in green and the worse set in blue
87
We can see that neither the blue nor green region forms a convex region in R2 . Take a line segment
on opposite sides of a corner in the blue region to disprove that the worse set is a convex region and
take the segment between the two split green regions to disprove that the better set is a convex region.
Since the function is neither quasiconvex nor quasiconcave then it is also neither convex nor concave.
3 −7/4 1/3 1 −3/4 −2/3
− x1 x2 x
3 1
x2
3. Compute the Hessian to get ∇2 π = 1 4 8 1/4 −5/3 . We start by testing it for strict
−3/4 −2/3
x x2 − x1 x2
3 1 9
definiteness. We compute the principle leading minors to be
3 −7/4 1/3
H1 = − x1 x2 < 0 as x1 , x2 > 0
4
3 −7/4 1/3 8 1/4 −5/3 1 −3/4 −2/3 1 −3/4 −2/3 5 −3/2 −4/3
H2 = − x1 x2 − x1 x2 − x1 x2 x1 x2 = x1 x2 >0
4 9 3 3 9
and thus by Sylvestor’s Criterion H is negative definite. As the Hessian is negative definite and the domain
is convex then the function π is strictly concave.
4 + 12x2
−4
4. The Hessian is given by H = . This is positive definite for all values except at (0, 0).
−4 4 + 12y 2
Thus it is strictly convex except at (0, 0). Since it’s strictly convex everywhere around that point it is also
convex there, thus the function is strictly convex.
5. Using the principle minors, the function is convex but not strict.
6. (a) Hessian is indefinite, function is neither convex nor concave.
(b) Hessian is negative semidefinite, function is concave.
(c) Hessian is negative definite, function is strictly concave.
(d) Hessian is positive definite, function is strictly convex.
(e) Hessian is positive definite, function is strictly convex.
88
7. Use the fact that ∇2 h(x) = a∇2 f (x) + b∇2 g(x) then argue that this is negative definite.
8. (a) The Hessian matrix is the zero matrix, thus you can’t use it to determine convexity or concavity.
However you may check using the definition that
f (tx + (1 − t)x0 , ty + (1 − t)y 0 ) = tf (x, y) + (1 − t)f (x0 , y 0 )
and thus it satisfies both convexity and concavity.
(b) The Hesian shows that is convex but not strict.
(c) The Hessian shows it is strictly concave.
(d) The Hessian shows it is strictly convex.
(e) The Hessian shows it is strictly convex.
8 −5/3 1/2 −2/3 −1/2
− x x2 2x1 x2
9. The Hessian is H = 3 1 which is negative definite using the method of minors.
−2/3 −1/2 1/3 −3/2
2x1 x2 −3x1 x2
Thus it is strictly concave.
6x1 + 4 2
10. The Hessian is H = which is positive semidefinite if 6x1 + 4 ≥ 0 and 6x1 ≥ 0. That is, it is
2 1
convex for x1 ≥ 0 and x2 can be any real number.
4 0 2
11. The Hessian is H = 0 0 2k . This is positive semidefinite if k = 0 and indefinite otherwise. Thus
2 2k 4
the function is only convex (but not strict) when k = 0.
12. The Hessian is
−2 2 0 0
2 −2 0 0
H=
0 0 −2 4
0 0 4 −2
The second order principal minor obtained by deleting the first and second rows and columns is -12. So the
matrix is not negative semidefinite and thus the function is not concave.
13. (a) It is convex in the region of all (x1 , x2 ) such that x1 ≥ x22 . It is never concave.
(b) It is convex in the region of all (x1 , x2 ) such that x1 ≥ 4. It is never concave.
(c) It is convex over the entire region of R2 . It is never concave.
(d) It is convex in the region of all (x1 , x2 ) such that 2x1 x2 < −1 and x1 > 0 or 2x1 x2 < −1 and x1 < 0.
The function is constant on the sets where x1 = 0 or x2 = 0 and thus convex and concave there. The
function is not concave anywhere else.
(e) The function is only defined if x1 , x2 > 0 or x1 , x2 < 0. The Hessian matrix is always negative definite
in these regions so it is concave. It is never convex.
89
6.5 Practice Questions
1. Determine, if possible, which of the following properties each of the following functions satisfies: convexity,
strict convexity, concavity, strict concavity, quasiconcavity, strict quasiconcavity.
(a) f (x, y) = x2 y 2
(b) f (x, y) = x − ex − ex+y
(c) f (x, y) = xy 2
(d) f (x, y) = 2x2 − xy + 2y 2
2. Determine whether or not each of the following functions is homogeneous, and if so of what degree?
(a) f (x, y) = 3x + 4y
(b) f (x, y) = 3x + 4y − 2
(c) f (x, y) = 2x2 + xy
(d) f (x) = x2 + x3
(e) f (x, y) = x1/2 y + x3/2
x3 − y 3
(f) f (x, y) = 1/2
x + y 1/2
(g) f (x, y) = x3 y 3 + x1/2
√ √ √
x+ y+ z
(h) f (x, y, z) =
x+y+z
x3
(i) f (x, y) = 30x1/2 y 3/2 − 2
y
(j) f (x1 , ..., xn ) = (g(x1 , ..., xn ))p where g is homogenous of degree n and p > 1.
3. For the functions of the previous exercise, show that each function that is homogeneous satisfies Euler’s
theorem. State whether they are increasing, descreasing or constant returns to scale.
4. Let f (x) be a non-constant positive and homogeneous function of degree k. Is there any such value of k so
that g(x) = ln(f (x)) is a homogeneous function?
g(x, y)
5. Let g(x, y) be a homogeneous function of degree 1. Is the function f (x, y) = k ln homogeneous?
x
6. Consider the following problems
(a) A firm’s differentiable production function f (x) is homogeneous of degree 1. Assume that the price (in
terms of output) of each input is its marginal product fi (x). What is the relation between the total
cost of the firm’s inputs and its output?
(b) A consumer’s differentiable demand function for some good if f (p, w) = f (p1 , p2 , ..., pn , w) where pi is
the price of the ith good, and w is the consumer’s wealth. Assume this function is homogeneous of
degree 0. Is there any necessary relationship between p1 f1 (p1 , ..., pn , w) + · · · + pn fn (p1 , ..., pn , w) and
wfn+1 (p1 , ..., pn , w)?
7. Revisit the proof of Euler’s theorem. Tweak one part of the proof to further prove the fact that if a twice
differentiable function is of degree k then
xT ∇2 f (x)x = k(k − 1)f (x)
90
Answers
1. (a) The test for convexity and concavity fails. Using the bordered Hessian you may find that the function
is quasiconcave.
(b) The function is strictly concave and thus also quasiconcave.
(c) The function is quasiconcave using the bordered Hessian.
(d) The function is strictly convex and thus also quasiconvex.
2. (a) Homogeneous of degree 1
(b) Not homogeneous
(c) Homogeneous of degree 2
(d) Not homogeneous
(e) Homogeneous of degree 3/2
(f) Homogeneous of degree 5/2
(g) Not homogeneous
(h) Homogeneous of degree -1/2
(i) Homogeneous of degree 2
(j) Homogenous of degree np
3. For Euler’s Formula, this is plug and play into a formula. You’ll know if you performed it correctly because
both sides of the equation should equal.
(a) Constant returns
(b) Does not apply
(c) Increasing returns
(d) Does not apply
(e) Increasing returns
(f) Increasing returns
(g) Does not apply
(h) Decreasing returns
(i) Increasing resturns
(j) Increasing as np > 1
4. You want the condition
ln(sk f (x)) = sm ln(f x)
to hold for some value of m. Use logarithmic properties to obtain ln(sk f (x)) = ln(sk ) + ln(f x) to obtain
ln(sk ) = (sm − 1) ln(f (x))
The equation will hold if both k = m = 0 leaving g to be homogeneous of degree 0. Otherwise if k 6= 0 6= m
then the right hand side will be a function of x while the left hand side is constant. This is impossible as
f (x) is not a constant function.
91
5. It is homogeneous of degree 0.
6. (a) The total cost of the firm’s inputs is
x1 f1 (x) + · · · + xn fn (x)
and the firm’s output if f (x). Since the function is homogeneous of degree 1 Euler’s theorem tells us
these are equal.
(b) By Euler’s theorem we have that p1 f1 (p1 , ..., pn , w) + · · · + pn fn (p1 , ..., pn , w) + wfn+1 (p1 , ..., pn , w) = 0
as f is of degree zero.
7. If you’re comfortable with the second order chain rule this shouldn’t be too bad. We have that at one point
in the proof
x1 fu1 (u) + · · · + xn fun (u) = ksk−1 f (x)
where ui = sxi for i = 1, ..., n. Take the derivative again with respect to s and observe that
d
[x1 fu1 (u)] = x1 (x1 fu1 u1 (u) + · · · xn fu1 un (u))
ds
Do this with each term and get a quadratic form. Meanwhile the right hand side will become k(k−1)sk−2 f (x).
Set s = 1 in both sides to obtain the result.
92
6.6 Practice Questions
1. Find the second order Taylor polynomial of f (x) = 1 + x + x2 at a = −1.
2. Find the third order Taylor polynomial of f (x) = e5x at a = 0.
3. Find the second order polynomial of f (x) = x1/3 at a = 8 and use it to approximate 111/3 .
√
4. Find the fourth order Taylor polynomial of f (x) = x at a = 4.
5. Suppose that an unknown function y = f (x) satisfies the two following properties
• y(0) = 1; and
• y 0 (x) = (y(x))2
use this to compute P3 (x) of this function centered at a = 0.
6. Determine the second degree Taylor polynomial approximation of the function f (x1 , x2 ) = x1 ex2 + 1 centered
at (1, 0) and use this to approximate f (1.1, −0.2).
√
7. Determine the second degree Taylor polynomial approximation of the function f (x1 , x2 ) = x1 x2 centered
at (1, 4) and use this to approximate f (0.9, 4.2).
8. Find the second order Taylor polynomial for the following functions about the given point.
(a) f (x1 , x2 ) = ln(1 + x1 + 2x2 ) about (2, 1)
(b) f (x1 , x2 ) = x31 + 3x21 x2 + 6x1 x22 − 5x21 + 3x1 x22 about the point (1, 2)
(c) f (x1 , x2 ) = ex1 +x2 about the point (0, 0)
(d) f (x1 , x2 , x3 ) = x1 − x22 + x1 x3 about the point (1, 0, 3)
93
Answers
1. f (x) = 1 − (x + 1) + (x + 1)2 .
25 2 125 3
2. P3 (x) = 1 + 5x + x + x
2 6
1 1
3. P2 (x) = 2 + (x − 8) − (x − 8)2 . Here 111/3 = f (11) ≈ P2 (11) = 2.25 Compared to the actual value of
12 288
111/3 ≈ 2.21875 this is fairly close.
1 1 1 5
4. P4 (x) = 2 + (x − 4) − (x − 4)2 + (x − 4)3 − (x − 4)4 .
4 64 512 16384
5. By plugging x = 0 into the equation we obtain y 0 (0) = (y(0))2 = (1)2 = 1. Thus y 0 (0) = 1. Differentiating
both sides of y 0 (x) = (y(x))2 and using the chain rule gives
d 0 d
y 00 (x) = (y(x))2 = 2y(x)y 0 (x)
[y (x)] =
dx dx
then plugging in x = 0 gives y 00 (0) = 2y(0)y 0 (0) = 2(1)(1) = 2. Repeat this procedure of differentiating the
equation and plugging in x = 0 to obtain the full results of y(0) = 1, y 0 (0) = 1, y 00 (0) = 2, y 000 (0) = 6. This
gives us
P3 (x) = 1 + x + x2 + x3
6. One may compute everything to obtain
x1 − 1
1 0 1 x1 − 1
f (x) ≈ 2 + 1 1 + x1 − 1 x2
x2
2 1 1 x2
1
= 2 + (x1 − 1) + x2 + (x1 − 1)x2 + x22
2
1
and we compute f (1.1, −0.2) ≈ 2 + (0.1) − 0.2 + (0.1)(−0.2) + (−0.2)2 = 1.9. Compared to the true value
2
of f (1.1, −0.2) = 1.9000603....
7. One may compute everything to obtain
x1 − 1
1 0 1/4 x1 − 1
f (x) ≈ 2 + 1 1/4 + x1 − 1 x2 − 4
x2 − 4
2 1/4 −1/32 x2 − 4
1 1 1
= 2 + (x1 − 1) + (x2 − 4) + (x1 − 1)(x2 − 4) − (x2 − 4)2
4 4 64
1 1 1
and we compute f (0.9, 4.2) ≈ 2 + (−0.1) + (0.2) + (−0.1)(0.2) − (0.2)2 = 1.944375. Compared to the
4 4 64
true value of f (0.9, 4.2) = 1.8444511...
1 2 1 2 2
8. f (x) ≈ ln(5) + (x1 − 2) + (x2 − 1) − (x1 − 2)2 − (x1 − 2)(x2 − 1) − (x2 − 1)2
5 5 50 25 25
9. f (x) ≈ 38 + 41(x1 − 1) + 39(x2 − 2) + 4(x1 − 1)2 + 42(x1 − 1)(x2 − 2) + 9(x2 − 2)2
94
1 1
10. f (x) ≈ 1 + x1 + x2 + x21 + x1 x2 + x22
2 2
11. f (x) ≈ 4 + 4(x1 − 1) + (x3 − 3) − x22 + (x1 − 1)(x3 − 3)
95
Chapter 7
Optimization of Functions of n-Variables
96
7.1 Practice Questions
1. Find all critical points of the following functions
(a) f (x1 , x2 ) = x21 − 2x1 x2 + 2x22 + 2x1 − 6x2 + 12
(b) f (x1 , x2 ) = 2x31 − 6x1 x2 + x22 + 4x2
(c) f (x1 , x2 ) = x1 x2 (5x1 + x2 − 15)
(d) f (x1 , x2 ) = x21 x2 + 2x22 − 2x1 x2 + 6
2 2
(e) f (x1 , x2 ) = e2x1 +x2
(f) f (x1 , x2 ) = (3x1 + 4x31 )(x22 + 2x2 )
2 2
(g) f (x1 , x2 ) = e−x1 −x2 +2x1 −2x2 −2
1
(h) f (x1 , x2 ) = x21 + x32 − 3x1 x2 − 4x1 + 2
2
(i) f (x1 , x2 ) = x31 + x32 + 2x1 + 6x2
(j) f (x1 , x2 ) = x41 + x42 − 4x1 x2 + 2
2 2
(k) f (x1 , x2 ) = (x21 + x22 )e−x1 −x2
(l) f (x1 , x2 ) = r ln(x2 ) − sx2 + t ln(x1 ) − ux1
(m) f (x1 , x2 ) = x31 + x32 + 3x21 − 3x22 − 8
x1 + x2
2. Find all critical points of the function f (x1 , x2 ) =
1 + x21 + x22
Hint: When forming f1 = 0 and f2 = 0 it will be easier to solve f1 + f2 = 0 and f1 − f2 = 0 (adding and
subtracting the equations). Then check this solves the original equations.
3. Find all critical points of the function f (x1 , x2 , x3 ) = x31 − 2x21 + x22 + x33 − 2x1 x2 + x1 x3 − x2 x3 + x3 .
Hint: Solve f2 = 0 and f3 = 0 for x1 and use this to solve f1 = 0.
4. Find all critical points of the function f (x1 , x2 , x3 ) = x31 + x32 + x23 + 2x1 x2 x3
5. Find all critical points of the function f (x1 , x2 , x3 ) = x41 − 5x21 x2 x3 + x22 + x23
6. Suppose that a firm produces two goods where they charge $1000 for good 1 and $800 for good 2. Let xi
denote the quantity of good i for i = 1, 2 and suppose the cost is given by C(x1 , x2 ) = 2x21 + 2x1 x2 + x22 .
(a) Determine the profit function of this firm.
(b) Determine all critical points of the profit function.
7. A firm is selling its product in two markets. In market A the demand is given by x1 = 100 − 2p1 and in
1 2
market B the demand is x2 = 80 − 4p2 . The firm’s total cost is C = x where x = x1 + x2 is the total
12
output.
(a) Construct the firms profit function as a function of x1 and x2 .
(b) Find the quantities and profits associated to the critical points of the profit function.
97
8. A firm charges different prices for its goods to its domestic and export markets. The inverse demand equation
3
for the domestic market is given by p1 = 500 − x1 and for the export market is given by p2 = 360 − x2 .
2
The total cost function is given by C(x1 , x2 ) = 50, 000 + 20x where x = x1 + x2 is the total output.
(a) Determine whether the profit function is convex, concave or neither.
(b) Use your previous result to find the prices the firm should charge to maximize profits.
9. In a certain community, there are two breweries in competition, so that sales of each negatively affect the
profits of the other. If brewery A produces x1 litres of beer per month and brewery B produces x2 litres per
month, then the profits of the two breweries are
2x21 + x22 4x22 + x21
PA = 2x1 − PB = 2x2 −
106 2 × 106
respectively. Find the sum of the two profits if each brewery independently sets its own production level to
maximize its own profit and summes that its competitor does likewise. Find the sum of the two profits if
the two breweries cooperate so as to maximize that sum.
98
Answers
1. (a) (1, 2)
(b) (1, 1) and (2, 4)
(c) (0, 0), (3, 0), (0, 15), (1, 5)
(d) (0, 0), (2, 0), and (1, 1/4)
(e) (0, 0)
(f) (0, 0) and (0, −2)
(g) (1, −1)
(h) (16, 4) and (1, −1)
(i) No critical points
(j) (0, 0), (1, 1) and (−1, −1)
(k) (0, 0) and all points (x1 , x2 ) such that x21 + x22 = 1
(l) (t/u, r/s)
(m) (0, 0), (−2, 0), (0, 2), and (−2, 2)
p p p p
2. ( 1/2, 1/2) and (− 1/2, − 1/2)
3. (0, −1, −2) and (2, 1, −2)
4. Just (0, 0, 0)
p p p p
5. (0, 0, 0), ( 2/5, 2/5, 2/5), (− 2/5, 2/5, 2/5), ( 2/5, −2/5, −2/5) and (− 2/5, −2/5, −2/5).
6. (a) π(x1 , x2 ) = R(x1 , x2 ) − C(x1 , x2 ) = 1000x1 + 800x2 − 2x21 − 2x1 x2 − x22
(b) (100, 300)
1 1 1
7. (a) π(x1 , x2 ) = 50 − x1 x1 + 20 − x2 x2 − (x1 + x2 )2
2 4 2
∗ ∗
(b) (x1 , x2 ) = (40, 20) and (p1 , p2 ) = (30, 15)
2 −2 0
8. (a) As ∇ π = it’s eigenvalues are λ1 = −2 < 0 and λ2 = −3 < 0. Thus it is strictly concave.
0 −3
(b) We obtain (x∗1 , x∗2 ) = (240, 340/3) and thus the prices are (p1 , p2 ) = (260, 570/3). This is a maximum
as the profit function is strictly concave (bowl shape down).
1 1
9. Set (PA )x1 = 0 and (PB )x2 = 0 to obtain x1 = 106 and 106 . Then we also find
2 2
1 5 2 2
PA + PB = 2(x1 + x2 ) − 6 x + 3x2
10 2 1
5
⇒ PA (106 , 106 ) + PB (106 , 106 ) = 106
8
1 5 2 2
On the other hand if (A, B) adjust (x1 , x2 ) to maximize P := PA + PB = 2(x1 + x2 ) − 6 x + 3x2 .
10 2 1
2 1
We solve Px1 = 0 and Px2 = 0 to obtain x1 = 106 and x2 = 106 . The value of P at this point is
5 3
2 6 1 6 11 6
P 10 , 10 = 10 which is greater than the previous result.
5 3 15
99
7.2 Practice Questions
1. Determine whether the following functions are convex or concave and use this to classify their critical point
as an absolute maximum or absolute minimum. If it is neither convex nor concave use further analysis to
determine if its a local maximum, local minimum, or a saddle point.
(a) f (x1 , x2 ) = x21 − 6x1 + x22 + 10x2 + 20
(b) f (x1 , x2 ) = 12 − 3x21 − 6x1 − x22 + 12x2
(c) f (x1 , x2 ) = x21 + 8x1 − 2x22 + 16x2
(d) f (x1 , x2 ) = x21 + 6x1 x2 + x22
2. Find and classify all critical points of the following functions
(a) f (x1 , x2 ) = x21 + x1 x2 + x22 + 3x1 − 3x2 + 4
(b) f (x1 , x2 ) = x21 + x1 x2 + 3x1 + 2x2 + 5
(c) f (x1 , x2 ) = 2x1 x2 − x21 − 2x22 + 3x1 + 4
(d) f (x1 , x2 ) = 2x21 + 3x1 x2 + 4x22 − 2x1 + 2x2 + 1
(e) f (x1 , x2 ) = x21 − x22 − 2x1 + 4x2 + 6
q
(f) f (x1 , x2 ) = 56x21 − 8x22 − 16x1 − 31 + 1 − 8x1
(g) f (x1 , x2 ) = x31 − x32 − 2x1 x2 + 6
(h) f (x1 , x2 ) = 6x21 − 2x31 + 3x22 + 6x1 x2
(i) f (x1 , x2 ) = x31 + 3x1 x22 − 15x1 + x32 − 15x2
(j) f (x1 , x2 ) = 4x1 x2 − x41 − x42
1
(k) f (x1 , x2 ) = 2
x1 + x22 − 1
2 2
(l) f (x1 , x2 ) = ex1 +x2 −4x1
(m) f (x1 , x2 ) = e−x2 (x21 + x22 )
(n) f (x1 , x2 ) = 2 ln(x1 ) + ln(x2 ) − 4x1 − x2
(o) f (x1 , x2 ) = 3x21 x2 + x32 − 3x21 − 3x22 + 2
3. Let f (x1 , x2 ) = x21 + kx1 x2 + x22 . For what values of k will the origin be a local minimum? For what values
of k will the origin be a saddle point?
4. Find and classify the critical point of f (x1 , x2 , x3 ) = x21 x2 − 2x1 x2 + x21 − 2x1 − x2 x3 . You will note that
local analysis fails when using Sylvester’s criterion for strict definiteness. Instead examine the eigenvalues.
5. Find and classify the critical point of f (x1 , x2 , x3 ) = x21 x2 − x22 + x23 + x2 x3 . You will note that local analysis
fails when using Sylvester’s criterion for strict definiteness. Instead examine the eigenvalues.
6. Suppose that f (x1 , x2 , x3 , x4 ) is a twice differentiable function with two critical point a and b and furthermore
you have found that
5 −2 3 1 1 2 3 12
−2 4 −1 0 2 4 1 2
∇2 f (a) = ∇2 f (b) =
3 −1 2 1 3 1 −2 5
1 0 1 7 12 2 5 3
100
If possible, classify these critical points.
7. In an earlier example it was established that a function of the form
1
f (x) = c + aT x + xT Cx
2
has a gradient given by ∇f (x) = a+Cx and Hessian ∇2 f (x) = C. Thus the first order conditions ∇f (x) = 0
gives x∗ = −C −1 a. Use this to find and classify the critical points of the following functions.
(a) f (x1 , x2 ) = 3x21 + 4x1 x2 + 9x22 + 3x1 − 2x2 + 7
(b) f (x1 , x2 ) = −2x21 − 5x22 + 2x1 − x2 + 10
101
Answers
1. (a) Convex, (3, −5) is an absolute minimum
(b) Concave, (−1, 6) is an absolute maximum
(c) It is neither convex nor concave, (−4, 4) is a saddle point
(d) It is neither convex nor concave, (0, 0) is a saddle point
2. (a) (−3, 3) is a local minimum
(b) (−2, 1) is a saddle point
(c) (3, 3/2) is a local maximum
(d) (2, −1) is a local minimum
(e) (1, 2) is a saddle point
(f) (16/7, 0) is a local maximum
(g) (0, 0) is a saddle point, (−2/3, 2/3) is a local maximum
(h) (0, 0) is a local minimum, (1, −1) is a saddle point
√ √
(i) (0, 5) and (0, − 5) are saddle points, (−2, −1) is a local maximum and (2, 1) is a local minimum
(j) (0, 0) is a saddle point, (1, 1) and (−1, −1) are local maxima
(k) (0, 0) is a local maximum
(l) (2, 0) is a local minimum
(m) (0, 0) is a local minimum and (0, 2) is a saddle point
(n) (1/2, 1) is a local maximum
(o) (0, 0) is a relative maximum, (0, 2) is a relative minimum, (1, 1) and (−1, 1) are saddle points.
3. (0, 0) is a local minimum for −2 < k < 2. (0, 0) is a saddle point for k > 2 or k < −2.
4. The only critical point is (1, 0, −1). The Hessian is
2x2 + 2 2x1 − 2 0 2 0 0
∇2 f (x) = 2x1 − 2 0 −1 ⇒ ∇2 f (1, 0, −1) = 0 0 −1
0 −1 0 0 −1 0
The eigenvalues may be found to be λ1 = −1 < 0, λ2 = 1 > 0 and λ3 = 2 > 0. As some eigenvalues have
differing signs the matrix is indefinite. Thus this critical point is a saddle point.
5. The only critical point is (0, 0, 0). The Hessian is
2x2 2x1 0 0 0 0
∇2 f (x) = 2x1 −2 1 ⇒ ∇2 f (0) = 0 −2 1
0 1 2 0 1 2
√ √
The eigenvalues of this are λ1 = − 5, λ2 = 0 and λ3 = 5. As some of the eigenvalues are differing in sign
then the matrix is indefinite. Thus this critical point is a saddle point.
6. a corresponds to a minimum and b corresponds to a saddle point.
102
3 6 4 ∗ −1 −16/23
7. (a) a = and C = so x = −C a = and as C 0 the point corresponds to
−2 4 8 6/23
a minimum.
2 −4 0 ∗ −1 1/2
(b) a = and C = so x = −C a = and as C ≺ 0 the points corresponds
−1 0 −10 −1/10
to a maximum.
103
Chapter 8
Constrained Optimization
104
8.1 Practice Questions
1. Find the spatial critical points (x∗1 , x∗2 ) of the Lagrange system associated to each of the following constraint
problems.
(a) Optimize f (x1 , x2 ) = x1 x2 subject to x21 + 2x22 = 1
(b) Optimize f (x1 , x2 ) = 49 − x21 − x22 subject to x1 + 3x2 = 10
(c) Optimize f (x1 , x2 ) = x21 + x22 subject to x1 x22 = 54
(d) Optimize f (x1 , x2 ) = x1 + x2 subject to x1 x2 = 16 where (x1 , x2 ) ∈ R2+
(e) Optimize f (x1 , x2 ) = x1 x2 subject to x1 + x2 = 16
(f) Optimize f (x1 , x2 ) = x1 x2 subject to x21 /16 + x22 /9 = 1
(g) Optimize f (x1 , x2 ) = x21 + x22 subject to x21 − 2x1 + x22 − 4x2 = 0
(h) Optimize f (x1 , x2 ) = 4x21 − 4x1 x2 + x22 subject to x21 + x22 = 25
2. Find all critical points (x∗1 , x∗2 , λ∗ ) of the Lagrange system associated to optimizing f (x1 , x2 ) = 4x21 + 3x1 x2 +
6x22 subject to x1 + x2 = 56.
3. A firm produces two goods x1 and x2 . Due to a government quota, the firm must produce subject to the
constraint x1 + x2 = 42. Find the critical point of optimizing the cost function c(x1 , x2 ) = 8x21 − x1 x2 + 12x22
subject to the given constraint.
4. Find all critical points (x∗1 , x∗2 , λ∗ ) of the Lagrange system associated to optimizing f (x1 , x2 ) = x21 + x22
subject to x21 + x1 x2 + x22 = 3.
5. Find all critical points (x∗1 , x∗2 , λ∗ ) of the Lagrange system associated to optimizing f (x1 , x2 ) = x1 x2 subject
to x21 + x22 = 2c2 where c > 0 is a constant.
6. Find all critical points (x∗1 , x∗2 , x∗3 , λ∗1 , λ∗2 ) of the Lagrange system associated to optimizing f (x1 , x2 , x3 ) =
x21 + x22 + x23 subject to 2x1 + x2 + 2x3 = 9 and 5x1 + 5x2 + 7x3 = 29
7. Find all critical points (x∗1 , x∗2 , x∗3 , λ∗1 , λ∗2 ) of the Lagrange system associated to optimizing f (x1 , x2 , x3 ) = x1
subject to x1 + x2 − x3 = 0 and x21 + 2x22 + 2x23 = 8
8. Find all critical points (x∗1 , x∗2 , x∗3 , λ∗1 , λ∗2 ) of the Lagrange system associated to optimizing f (x1 , x2 , x3 ) =
4 − x3 subject to x21 + x22 = 8 and x1 + x2 + x3 = 1.
9. Find all the critical points (x∗1 , x∗2 , λ∗ ) of the Lagrange system associated to optimizing f (x1 , x2 ) = x21 + x22
subject to x21 + x2 = 1.
10. Find all critical points (x∗1 , x∗2 , λ∗ ) of the Lagrange system associated to optimizing f (x1 , x2 ) = x21 + 2x22 − x1
subject to x21 + x22 = 1.
11. Find all spatial critical points (x∗1 , x∗2 , x∗3 ) of the Lagrange system associated to optimizing f (x1 , x2 , x3 ) =
x21 + x22 + x23 subject to x23 = x21 + x22 and x1 + x2 − x3 + 1 = 0
12. We know that the Cobb-Douglas function f (x1 , x2 ) = Axa1 xb2 is concave for A > 0, 0 < a, b < 1 with
a + b ≤ 1 and thus any critical point obtained will be a maximum. Suppose a company has determined that
5 9/20 11/20
its production level is given by the Cobb-Douglas function f (x1 , x2 ) = x1 x2 where x1 represents the
2
total number of labour hours in 1 year and x2 represents the total capital input for the company. Suppose
1 unit of labor costs $40 and 1 unit of capital costs $50. Find the maximum value of this function subject
to a budgetary constraint of $500, 000 per year.
105
Answers
√ √ √ √
1. (a) (1/ 2, 1/2), (−1/ 2, 1/2), (1/ 2, −1/2) and (−1/ 2, −1/2)
(b) (1, 3)
√ √
(c) (3, 3 2) and (3, −3 2)
(d) (4, 4) and (−4, −4)
(e) (8, 8)
√ √ √ √ √ √ √ √
(f) (2 2, 3 2/2), (2 2, −3 2/2), (−2 2, 3 2/2) and (−2 2, −3 2/2)
(g) (0, 0) and (2, 4)
√ √ √ √ √ √ √ √
(h) ( 5, 2 5), (− 5, −2 5), (−2 5, 5) and (2 5, − 5)
2. (x1 , x2 , λ) = (36, 20, 348)
3. (x1 , x2 , λ) = (25, 17, 383)
√ √ √ √
4. (x1 , x2 , λ) = (1, 1, 2/3), (−1, −1, 2/3), ( 3, − 3, 2) and (− 3, 3, 2)
5. (x1 , x2 , λ) = (c, c, 1/2), (−c, −c, 1/2), (c, −c, −1/2) and (−c, c, −1/2)
6. (x1 , x2 , x3 , λ1 , λ2 ) = (2, 1, 2, 2, 0)
7. (x1 , x2 , x3 , λ1 , λ2 ) = (2, −1, 1, 1/2, 1/8) and (−2, 1, −1, 1/2, −1/8)
8. (x1 , x2 , x3 , λ1 , λ2 ) = (2, 2, −3, 1/4, −1) and (−2, −2, 5, −1/4, −1)
9. (x1 , x2 , λ) = (0, 1, 2), (1, 0, 0) and (−1, 0, 0)
√ √
10. (x1 , x2 , λ) = (1, 0, 1/2), (−1, 0, 3/2), (1/2, 3/2, 0) and (1/2, − 3/2, 0)
√ √ ! √ √ !
2 2 √ 2 2 √
11. −1 + , −1 + , −1 + 2 and −1 − , −1 − , −1 − 2
2 2 2 2
12. Subject to the given constraint, a maximum production level of 13890 occurs with 5625 labor hours and
$5500 of total capital input.
106
8.2 Practice Questions
1. Classify all the critical points of Question #2 of the previous section.
2. Classify all the critical points of Question #3 of the previous section.
3. Classify all the critical points of Question #4 of the previous section.
4. Classify all the critical points of Question #5 of the previous section.
5. Classify all the critical points of Question #6 of the previous section.
6. Classify all the critical points of Question #7 of the previous section.
7. Classify all the critical points of Question #8 of the previous section.
107
Answers
1. One may compute
0 −1 −1
∇2 L(x, λ) = −1 8 3
−1 3 12
which is constant at all values. As there is only one constraint then we need only determine the sign of
|H3 | = |H|. Computing the determinant gives |H| = −14 < 0 so it corresponds to a local minimum.
2. One may compute
0 −1 −1
∇2 L(x, λ) = −1 16 −1
−1 −1 24
which is constant at all values. As there is only one constraint then we need only determine the sign of
|H3 | = |H|. Computing the determinant gives |H| = −42 < 0 so it corresponds to a local minimum.
3. One may compute
0 −(2x1 + x2 ) −(x1 + 2x2 )
∇2 L(x, λ) = −(2x1 + x2 ) 2 − 2λ −λ
−(x1 + 2x2 ) −λ 2 − 2λ
As there is only one constraint then we need only determine the sign of |H3 | = |H| for each point.
• At (1, 1, 2/3):
0 −3 −3
∇2 L(1, 1, 2/3) = −3 2/3 −2/3 ⇒ |H| = −24 < 0
−3 −2/3 2/3
and thus the point corresponds to a minimum.
• At (−1, −1, 2/3):
0 3 3
∇2 L(−1, −1, 2/3) = 3 2/3 −2/3 ⇒ |H| = −24 < 0
3 −2/3 2/3
and thus the point corresponds to a minimum.
√ √
• At ( 3, − 3, 2):
√ √
√ √ √0 − 3 3
∇2 L( 3, − 3, 2) = −√ 3 −2 −2 ⇒ |H| = 24 > 0
3 −2 −2
and thus the point corresponds to a maximum.
108
√ √
• At (− 3, 3, 2):
√ √
√ √ √0 3 − 3
∇2 L(− 3, 3, 2) = √3 2/3 −2/3 ⇒ |H| = 24 > 0
− 3 −2/3 2/3
and thus the point corresponds to a maximum.
4. One may compute
0 −2x1 −2x2
∇2 L(x, λ) = −2x1 −2λ 1
−2x2 1 −2λ
As there is only one constraint then we need only determine the sign of |H3 | = |H| for each point.
• At (c, c, 1/2):
0 −2c −2c
∇2 L(c, c, 1/2) = −2c −1 1 ⇒ |H| = 16c2 > 0
−2c 1 −1
and thus the point corresponds to a minimum.
• At (−c, −c, 1/2):
0 2c 2c
∇2 L(−c, −c, 1/2) = 2c −1 1 ⇒ |H| = 16c2 > 0
2c 1 −1
and thus the point corresponds to a minimum.
• At (c, −c, −1/2):
0 −2c 2c
∇2 L(c, −c, −1/2) = −2c 1 1 ⇒ |H| = −16c2 < 0
2c 1 1
and thus the point corresponds to a maximum.
• At (−c, c, −1/2):
0 2c −2c
∇2 L(−c, c, −1/2) = 2c 1 1 ⇒ |H| = −16c2 < 0
−2c 1 1
and thus the point corresponds to a maximum.
5. One may compute
0 0 −2 −1 −2
0 0 −5 −5 −7
∇2 L(x, λ) =
−2 −5 2 0 0
−1 −5 0 2 0
−2 −7 0 0 2
109
which is constant at all values. As there is two constraints with three variables we need to compute |Hj |
for 2(2) + 1 ≤ j ≤ 2 + 3 ⇒ 5 ≤ j ≤ 5 ⇒ j = 5. That is, we only need to compute |H5 | = |H| (the whole
determinant yet again). Computing the determinant gives |H| = 100 > 0 and thus the point corresponds to
a local minimum (In case you just became confused... reminder that the signs you consider for
|Hj | change conditions depending on the number of variables and constraints).
6. One may compute
0 0 −1 −1 1
0 0 −2x1 −4x2 −4x3
∇2 L(x, λ) =
−1 −2x1 −2λ2 0 0
−1 −4x2 0 −4λ2 0
1 −4x3 0 0 −4λ2
Like the previous example we must compute |H5 | = |H| at all critical points.
• At (2, −1, 1, 1/2, 1/8):
0 0 −1 −1 1
0 0 −4 4 −4
∇2 L(2, −1, 1, 1/2, 1/8) =
−1 −4 −1/4 0 0 ⇒ |H| = −64 < 0
−1 4 0 −1/2 0
1 −4 0 0 −1/2
Thus the point corresponds to a maximum.
• At (−2, 1, −1, 1/2, −1/8):
0 0 −1 −1 1
0 0 4 −4 4
∇2 L(−2, 1, −1, 1/2, −1/8) =
−1 4 1/4 0 0 ⇒ |H| = 64 > 0
−1 −4 0 1/2 0
1 4 0 0 1/2
Thus the point corresponds to a minimum.
7. One may compute
0 0 −2x1 −2x2 0
0 0 −1 −1 −1
∇2 L(x, λ) =
−2x1 −1 −2λ1 0 0
−2x2 −1 0 −2λ1 0
0 −1 0 0 0
Like the previous example we must compute |H5 | = |H| at all critical points.
• At (2, 2, −3, 1/4, −1):
110
0 0 −4 −4 0
0 0 −1 −1 −1
∇2 L(2, 2, −3, 1/4, −1) =
−4 −1 −1/2 0 ⇒ |H| = −16 < 0
0
−4 −1 0 −1/2 0
0 −1 0 0 0
Thus the point corresponds to a maximum.
• At (−2, −2, 5, −1/4, −1):
0 0 4 4 0
0 0 −1 −1 −1
∇2 L(x, λ) =
4 −1 1/2 0 ⇒ |H| = 16 > 0
0
4 −1 0 1/2 0
0 −1 0 0 0
Thus the point corresponds to a minimum.
111
You might also like
- Mathematics For Economists by Carl P. Simon and Lawrence E. Blume (2004)92% (24)Mathematics For Economists by Carl P. Simon and Lawrence E. Blume (2004)952 pages
- Carl P. Simon, Lawrence E. Blume - Mathematics For Economists-W. W. Norton & Company (1994)100% (2)Carl P. Simon, Lawrence E. Blume - Mathematics For Economists-W. W. Norton & Company (1994)953 pages
- Mathematics For Economists by Carl P Simon and Lawrence E Blume 2004 PDF100% (1)Mathematics For Economists by Carl P Simon and Lawrence E Blume 2004 PDF952 pages
- Student's Manual Essential Mathematics For Economic Analysis100% (2)Student's Manual Essential Mathematics For Economic Analysis100 pages
- Applied Linear Algebra and Optimization Using MATLABNo ratings yetApplied Linear Algebra and Optimization Using MATLAB1,176 pages
- Further Mathematics For Economic Analysi PDFNo ratings yetFurther Mathematics For Economic Analysi PDF310 pages
- Introduction To Mathematical Economics Part 1100% (1)Introduction To Mathematical Economics Part 1271 pages
- Components, Moment and Resultant of Spatial ForcesNo ratings yetComponents, Moment and Resultant of Spatial Forces10 pages
- S0lution-M@+nu@l SydsaeterhammondarnestromNo ratings yetS0lution-M@+nu@l Sydsaeterhammondarnestrom77 pages
- Takashi's Econ633 Lecture Notes May 18 2010No ratings yetTakashi's Econ633 Lecture Notes May 18 2010105 pages
- Mathematical Economics Lecture Notes: Alexander W. RichterNo ratings yetMathematical Economics Lecture Notes: Alexander W. Richter128 pages
- Basic Mathematics For Economic Analysis INo ratings yetBasic Mathematics For Economic Analysis I12 pages
- Lecture Notes: Guoqiang TIAN Department of Economics Texas A&M University College Station, Texas 77843 (Gtian@tamu - Edu)No ratings yetLecture Notes: Guoqiang TIAN Department of Economics Texas A&M University College Station, Texas 77843 (Gtian@tamu - Edu)218 pages
- Lecture Notes For ECON660 and ECON460-2022-08No ratings yetLecture Notes For ECON660 and ECON460-2022-08265 pages
- Lecture Notes in LInear Algebra - ICEF MoscowNo ratings yetLecture Notes in LInear Algebra - ICEF Moscow121 pages
- B.A. P Economics Basic Mathemat IEvciDd-1No ratings yetB.A. P Economics Basic Mathemat IEvciDd-112 pages
- Mathematics For Business Economics: Herbert Hamers, Bob Kaper, John KleppeNo ratings yetMathematics For Business Economics: Herbert Hamers, Bob Kaper, John Kleppe37 pages
- Math 111f College and Advanced Algebra LATEST2024No ratings yetMath 111f College and Advanced Algebra LATEST2024116 pages
- Mathematics For Economists by Carl P Simon and Lawrence E Blume 2004 PDFNo ratings yetMathematics For Economists by Carl P Simon and Lawrence E Blume 2004 PDF952 pages
- Mathematical Modeling and Applied Calculus Joel Kilty PDF DownloadNo ratings yetMathematical Modeling and Applied Calculus Joel Kilty PDF Download59 pages
- (Lecture Notes) Jürgen Müller-Linear Algebra (2018) PDFNo ratings yet(Lecture Notes) Jürgen Müller-Linear Algebra (2018) PDF139 pages
- Associative Algebras and Schur-Weyl Duality100% (1)Associative Algebras and Schur-Weyl Duality68 pages
- Derivation of The Time Domain Solution of State EquationsNo ratings yetDerivation of The Time Domain Solution of State Equations3 pages
- Uffah Cademy: Test Mcqs - Math Objective Part Class: 6 Q.1 Choose The Correct Answer and Circule The Right Option 10No ratings yetUffah Cademy: Test Mcqs - Math Objective Part Class: 6 Q.1 Choose The Correct Answer and Circule The Right Option 103 pages
- Lecture 17 Properties of Fourier TransformNo ratings yetLecture 17 Properties of Fourier Transform6 pages
- Trigonometry: by Faudhi Issack Phone +255 655 413 177No ratings yetTrigonometry: by Faudhi Issack Phone +255 655 413 17771 pages
- ICSE Class 10 Maths Sample Paper 3 20200No ratings yetICSE Class 10 Maths Sample Paper 3 202005 pages
- 08 Maths Key Notes CH 09 Algebraic Expressions and IdentitiesNo ratings yet08 Maths Key Notes CH 09 Algebraic Expressions and Identities2 pages
- (PreCal 11) Conic Section - Ellipse PDFNo ratings yet(PreCal 11) Conic Section - Ellipse PDF49 pages
