July 27, 2009

Mathematical Economics: 2009-8-17(Mon)–21(Fri): TF: Kagawa

1 What to Study
Mathematical Economics. . . A study of economics, using advanced mathematical models.
Cf. Economics. . .A study of a capitalist economy, or a group of economies: how equilibrium
states are determined, and how the government can shift an equilibrium position.
Cf. General Equilibrium Analysis: A method in the study of a national economy, deriving
macro-relations from basic behaviors of individual agents. Initiated by L.Walras.
Cf. Marxian Economics: A study on how a capitalist economy functions, and how it deteriorates
and breaks down in the long-run.

2 Warnings at the Beginning

Please note that we study economics, not mathematics. Math is a tool, and grasp it intuitively,
or geometrically. Draw diagrams, and do real calculations on your notebooks by yourself with
no help of computers.

3 Vector Spaces
Real numbers: this is a real vector space of one
 dimension,
 and written
  as R.
x1 y1
 x2   y2 
   
We consider an n-tuple of real numbers: x ≡  .. , y ≡  .. , and so on. It is sometimes
 .   . 
xn yn
convenient to write x (or y) as

x0 ≡ (x1, x2 , . . . , xn )
with ()0 meaning transposition. (NB. Vectors are represented in bold face.)
We define the sum, difference, scalar multiplication between two n-tuples.
Sum and difference: x0 ± y0 ≡ (x1 ± y1, x2 ± y2, . . . , xn ± yn ), in short, elementwise.
scalar multiplication: kx0 ≡ (kx1 , kx2 , . . . , kxn ), in short, all the elements. k ∈ R.
We also define the inner-product of two n-tuples:
X
n
p·x≡ pi · xi = p1 · x1 + p2 · x2 + · · · + pn · xn ,
i=1

where p = (p1 , p2, . . . , pn ) and x0 ≡ (x1 , x2, . . . , xn ). Often, p · x is written also as px.
(NB. p is a row vector, while x a column vector. Hereinafter, p and q show row vectors: normally
vectors are column ones.)
We then define the Euclidean distance between two n-tuples, x and y:
v
u n
p uX
kx − yk ≡ (x − y) · (x − y) = t (xi − yi )2 .
0

i=1

Def: With these operations, sum, difference, and scalar multiplication, the set of all real n-tuples
forms a vector space of n dimensions. With the inner-product and Euclidean distance defined,
the space is the Euclidean space of n dimensions.
 July 27, 2009

4 Convergence
A sequence of vectors is simply a map from N to Rn , and is denoted as {xn }n∈N , or simply {xn },
where N is the set of natural numbers. (What you have learned is a sequence of (real) numbers,
which are in R.)
Def. A vector sequence {xn } is said to converge to x∗ when for any small positive number ²,
there exists an m ∈ N such that kxn − x∗ k < ² for all n > m.

5 Mappings
A map (or a mapping), T , between two spaces, X and Y , is a rule concerning how a particular
element in one space, X, is moved (or mapped) to an element in the other space, Y , and written
as: T : X → Y , or T : x ∈ X → T (x) ∈ Y . A function is a map from Rn to R. (What you
learned as a function is from R to R.)
When there is only one x ∈ X such that T (x) = y for y ∈ T (X) ≡ {y | y = T (x) for some
x ∈ X}, T is said to be one-to-one between X and T (X).
Suppose that T is one-to-one between X and T (X), then we can define the inverse map T −1 :
T (X) → X.
Def. A function, f : Rn → R, is continuous, when {xn } → x∗ implies {f(xn )} → f(x∗ ).
Def. A map T from a vector space X to a vector space Y is called linear, if it satisfies two
conditions:
(i) [additivity] T (x1 + x2 ) = T (x1) + T (x2 ) for x1 , x2 ∈ X and
(ii)[homogeneity] T (kx) = kT (x) for k ∈ R and x ∈ X.

6 Differentiation
To find out a maximum or a minimum, we differentiate functions, if they are differentiable. (Most
functions you know are differentiable, no worry.)
Rules of differentiation:
(i) (f (x) + g(x))0 = f 0 (x) + g 0 (x); (ii) (k · f (x))0 = k · f 0 (x); (iii) (f (x) · g(x))0 = f 0 (x) · g(x) +
f(x) · g 0 (x);
(iv) (f(g(x)))0 = f 0 (u) · g 0 (x), [u = g(x)].
NB. The property (iv) is called “chain rule”.
Rx
(v) (xn )0 R= n · xn−1 ; (vi) (ex )0 = ex ; (vii) (loge x)0 = 1/x for x > 0; (viii) ( c f(t, x)dt)0 =
x
f(x, x) + c (∂f /∂x)dt.

Exercises:
Q01. What are the limits for the following sequences of numbers.
(i) (1 + n1 )n ; (ii) { sin(1/n)
1/n
}; (iii) (A · K −1/n + (1 − A) · L−1/n )−k··n .
f 0 (x)
NB. L’Hopital’s Rule: Suppose f(x) → 0, g(x) → 0, g 0 (x)
→ C ( and g 0 (x) 6= 0) as x → a, then
f (x)
g(x)
→ C as x → a.

Q02. Differentiate the following functions wrt. x.

(i) loga x; (ii) (x3 + 2x − 7)11 ; (iii) (loge x)3 .
 July 27, 2009

7 Optimization with a constraint

Maximization Problem with a constraint: maximize f(x) subject to g(x) = 0, where
f, g : Rn → R.
Lagrange Multiplier Method: Construct the Lagrangian function L ≡ f(x) − λ · g(x), and
solve the following system of simultaneous equations:
 ∂L
 ∂xi = 0, i = 1, 2, . . . n
 ∂L
∂λ
= 0.
There are n + 1 equations for n + 1 variables, x1 , x2 , . . . , xn , and λ.
NB. The last equation is the constraint itself.
NB. When conducting partial differentiation, regard the other variables as constants.

8 Theory of Consumers’ Demand

Consumers are supposed to maximize their “utility”, u(), subject to their respective budget
constraint:
maximize u(x) subject to p · x = E. Here E is a given positive amount of expenditures, and
p a given price vector.
NB. In this problem, our utility is simply a function of consumption basket x. Cf. Veblen effect.
NB. Never fail to transform the constraint into the form g(x) = 0.
1. the case of two commodities and u(x) = A · xα1 · xβ2 . (parameters: A, α, β > 0.)
First, form the Lagrangian L = A · xα1 · xβ2 − λ · (p1 · x1 + p2 · x2 − E). Then we get
 ∂L α−1 β
 ∂x1 = A · α · x1 · x2 − λ · p1 (1)
∂L
= A · β · xα1 · xβ−1
2 − λ · p2 (2)
 ∂x2
−(p1 · x1 + p2 · x2 − E) = 0 (3)
(1) × x1 ⇒ A · α · xα1 · xβ2 = λ · p1 · x1 (4)
(2) × x2 ⇒ A · β · xα1 · xβ2 = λ · p2 · x2 (5)
Dividing (4) by (5), we get
α p1 · x1
= .
β p2 · x2

Substitute p2 · x2 = βα · p1 · x1 into (3), and we have x1 = ( α+β

α
· E) / p1 . In a similar way, we
β
obtain x2 = ( α+β · E) / p2 .
In sum, divide the budget into two parts of the ratios, α to β. Spend these parts for commodities
1 and 2 respectively.
Q00. When p1 goes up, what will happen to the consumption basket?

Exercises:
Q01. Consider the case of two commodities and u(x) = αx1 + βx2, and derive the utility
maximizing consumption basket.
Q02. Consider the case of three commodities and u(x) = A · xα1 · xβ2 · xγ3 , and derive the utility
maximizing consumption basket.
 July 27, 2009

9 Theory of Consumers’ Demand (continued)

Def. An indifference curve is a curve each consumption basket on which gives the same utility
level.

(1) the case of Cobb=Douglas utility function (2 commodities): No Substitution

u(x, y) = A · xα · yβ ; prices px , py ; expenditures E. See “mathecon 02.pdf”
α E β E
x= · , y= ·
a + β px a + β py
Exercises:
Q01. When our utility function is of Cobb=Douglas type, what is the effects of the increase in
px on x and y? (No substitution at all!).
Q02. Let u(x, y) = x·y, px = 2, py = 3, and E = 12. Draw indifference curves (E = 6, 10, 12, 16)
and the area where the budget constraint is satisfied.

(2) the case of a linear utility function (2 commodities): u(x) = αx + βy; prices px , py ;
expenditures E. Sudden Perfect Substitution
Suppose you buy only x, then you can get pEx , securing the utility level α · pEx . On the other
hand, if you buy only y, then you can get pEy , securing the utility level β · pEy . Thus, when
α · pEx > β · pEy , you buy only x as much as pEx . When α · pEx < β · pEy , you buy only y as much
as pEy you can get pEy , securing the utility level β · pEy . When α · pEx = β · pEy , you may buy either
commodity, or mix in any ratio. So, as px changes from a very low price to a higher one, a
complete substitution takes place all of a sudden.
The area where the budget constraint, px · x + py · y ≤ E, is satisfied forms a triangle. However,
only the hypotenuse is relevant.
Exercises:
Q03. Consider the case of two commodities, and u(x) = 2x + 3y, px = 20, py = 100, and
E = 10000. Derive the utility maximizing consumption basket. Find out the price of x above
which y substitutes x.
Q04. Consider the case of three commodities, and u(x) = 2x + 3y + 4z, px = 20, py = 60, pz =
100 and E = 12000. Derive the utility maximizing consumption basket. Now suppose the price
of x starts to rise, and you wish to find out the price of x , above which x is substituted for by
another commodity. By which commodity, is x substituted?
Answer. If we buy x only, we can get E/px = 12000/20 = 600, thus giving us the utility level
u = 2 × 600 = 1200. Similarly, concentrating on y, we secure the utility level u = 3 × 200 = 600,
and by z, u = 480. Therefore, we buy only x in the amount of 600, with y = z = 0. As the price
of x, px, goes up, the quantity of x we can buy decreases, hence the utility level also decreases
by sticking to x. When px = 40, we can buy x as much as 300(= 12000/40), giving the utility
level 600(= 2 × 300), which is the same level if we concentrate on y. Hence, substituting y for x
takes place when px goes over 40.

(3) the case of a min(,) utility function (2 commodities): L-shaped Indifference Curves and
Complements
Now we consider the utility function u(x, y) = min(x/a, y/b), which attains the smaller value
between two arguments. In this case, indifference curves become L-shaped, and the highest
utility level is attained at the south-west corner of an L-shaped indifference curve, which corner
should be on the hypotenuse of the budget constraint triangle. (Draw a diagram.) Thus, as the
 July 27, 2009

price of x goes up, both commodities x and y decreases, and vice versa: they are complements.
These are like a desk & a chair, coffee & a coffee cup etc.
Exercises:
Q05. Let u(x, y) = min(x/3, y/4), px = 20, py = 60, and E = 12000. What is the consumption
basket which maximizes utility.
Answer. Suppose we buy 3k of x and 4k of y, then the expense is 20 × 3k + 60 × 4k, which
should be equal to E = 12000. Thus, 300k = 12000, getting k = 40. Hence, one should buy
x = 120 and y = 160, yielding utility 40.

Q06. Now take up the case of three commodities, and let u = min(x/3, y/4, z/5), px = 20,
py = 30, pz = 40, and E = 12000. Find out the consumption basket with the maximum utility.

10 Order
Def. An order is a binary relation ¹ on a given set which satisfies the following three properties.
(i) reflexivity: a ¹ a;
(ii) symmetry: If a ¹ b and b ¹ a, then a = b;
(iii) transitivity: If a ¹ b and b ¹ c, then a ¹ c.

Examples. (a) The set of natural numbers, N . (b) The set of integers, Z.
(c) The set of rational numbers, Q. (d) The set of real numbers, R.
(e) In the Euclidean space of Rn , we consider an order ¹ which is defined as x ¹ y iff xi ≤ yi
for all i. (that is, when a vector y is to the north-east of a vector x , we say y is greater than
or equal to x.
(e) In the commodity space of R2+ , the non-negative orthant of R2 , we consider an order ¹
which is defined as in the case of Rn .
Def. A set is called a partially ordered set when it has an order.
Def. A map T form an partially ordered set to another partially ordered set is called “isotone”,
if x ¹ y implies T (x) ¹ T (y).

11 Fixed Point Theorems

References:
Tom M. Apostol, Mathematical Analysis, 2nd ed., Addison Wesley (Menlo Park), 1974. *detailed
Alpha C. Chiang, Fundamental Methods of Mathematical Economics, 2nd ed.,
McGraw-Hill, 1967, 1974. (A translation from McGraw-Hill Shuppan, Tokyo.) * a starter
Robert Dorfman, Paul A. Samuelson, Robert M. Solow, Linear Programming and Economic
Analysis, McGraw-Hill (New York), 1958.
Kelvin Lancaster, Mathematical Economics, New York: Dover (New York), 1968. * a starter
William Novshek, Mathematics for Economists, Academic Press (San Diego), 1993. *not easy
Eugene Silberberg, The Structure of Economics, McGraw Hill (New York), 1990.
Akira Takayama, Mathematical Economics, Cambridge Univ Press,1985.

Def. A segment [x, z] in Rn is defined as the set {y | y ∈ Rn such that x ¹ y ¹ z}

 July 27, 2009

Def. In a partially ordered set Y , an upper bound u of a subset S ⊂ Y is an element such that
x ¹ u for ∀x ∈ S .
Def. The least upper bound of a subset S, supS, in a partially ordered set Y , is the minimum
element, if any, in the set of all upper bounds of S.
Tarski’s Fixed Point Theorem. If a map T from a segment, S, of Rn into itself is isotone,
it has at least one fixed point.
Proof. Consider the set B ≡ {x | x ¹ T x, x ∈ S}, and let the least upper bound of S be x∗ .
From x ¹ x∗ for ∀x ∈ S, we have x ¹ T x ¹ T x∗ by isotoneness and the fact x ∈ S. Thus, it
follows

x∗ ¹ T x∗ .

Again, by isotoneness, T x∗ ¹ T (T x∗ ), which shows that T x∗ ∈ S, therefore we get

T x∗ ¹ x∗ .

From the above two inequalities, we have the desired result, x∗ = T x∗ . 2

Def. A map T from a subset of Rn into Rn is said to be continuous , if a sequence {xn }

converges x∗ , then {T (xn )} converges to T (x∗ ).
Brouwer’s Fixed Point Theorem. If a map T from a segment, S, of Rn into itself is contin-
uous, it has at least one fixed point.
Proof. (Omitted, because we need many pages.) 2

Red Herring: A great many people regard the above theorem as “spine-tingling”. This says:
(1) When you stir coffee or tea in a cup quietly, making no splash, there is at least one molecule
of water which does not move. (Continuous, because you stir quietly with no drop of water
jumping out of the surface.) (The surface of coffee in a cup can be continuously transformed to
a segment in R2.)
(2) You spread a sheet of plastic wrap on the table, and then put it in your palm and crease it
without tearing. When you return that rucked piece where it was on the table in the beginning,
there is at least one molecule which is at the same place where it was.
Reading: Robert Louis Stevenson, The Bottle Imp, 1893.
(You can read at: http://gaslight.mtroyal.ab.ca/bottlimp.htm and many other sites.
By the way, Stevenson preferred gas light to electric one.)

12 Fixed Point Theorem (continued): Brouwer and Kaku-

tani
Def. A set which can be included in a ball of a sufficiently large radius is bounded.
Def. A set is closed iff any converging sequence in the set has a limit point also in it.
Def. A bounded and closed set is compact. (This is valid only in the space of a finite dimension.)
Brouwer’s Fixed Point Theorem can be generalized to a continuous map T from a compact
convex set into itself.

Def. A correspondence from X to Y is a map which assigns a subset in Y for a point in X.

Def. A correspondence f from X to Y is upper semi-continuous iff given any sequence in X,
{xn } → x∗ , yn ∈ f(xn ), and {yn } → y∗ imply y ∗ ∈ f (x∗ ).
 July 27, 2009

Kakutani’s Fixed Point Theorem: An upper semi-continuous correspondence from a com-

pact convex set into itself has at least one fixed point.

Application to Game Theory : n player non-cooperative games

Each player has a finite number of pure strategies, which she/he can mix with certain probabili-
ties. Then each player has a compact convex set of strategies. The direct product S of strategy
sets of all players also forms a compact convex set.
Now given a point in this S, players modify their strategy so as to maximize their pay-off. This
correspondence can be regarded as upper semi-continuous. Hence, there is a fixed point.
Existence of a Nash Equilibrium: In a finite game with mixed strategies, there is a Nash
equilibrium.
NB. John Nash proved the existence in 1950, using Kakutani’s fixed point theorem. He received
a Nobel Prize in 1994.

13 Production
Production : Supply = Demand : Consumption and Investment
Supply of commodities ⇐ profit maximization
Def. profit = sales - costs

Production function: F (x1 , x2 , . . . , xn , k1 , k2, . . . , km , `1, `2, . . . , `s );

xi : material inputs, kj : durable machines, `t : labor power
pi : price of material inputs, rj : rental of durable machines, wt : wage rate of labor power of
type t.
p = the price of product P P Ps
Profit = p·F (x1 , x2, . . . , xn , k1 , k2 , . . . , km , `1 , `2 , . . . , `s )−( ni=1 pi ·xi + m
j=1 rj ·kj + t=1 wt ·`t )

14 Framework of General Equilibrium Analysis

Economic Agents: Consumers, Workers, and Firms, all mixed: the government behind
the scene.
Behavioral Assumptions: Every agent tries to maximize her/his objective function in a
selfish way. Utility functions for consumers and workers, profit functions for firms.
Each agent decides her/his demand/supply independently others, taking prices as given.
Can there be an equilibrium, where every market sees its supply and demand in balance?
Can every market reach and stay at an equilibrium?
Three Classical Problems:
(1) Existence, (2) Uniqueness, and (3) Stability of Equilibrium.

15 Production Function
A typical production function for a firm is S-shaped: first exhibiting increasing returns to
scale(irs) and then when the scale of production reaches a certain level, showing decreasing
returns to scale(drs).

n
Def. A function f from R+ to R is positively homogeneous of degree r iff f (kx) = k r ·f(x)
n
for x ∈R+ , k ∈ R+ .
 July 27, 2009

Euler’s Theorem: If f is positively homogeneous of degree r and differentiable at x, then

X
n
∂f (x)
· xi = r · f(x).
i=1
∂xi

Proof. Differentiate f (kx) = k r · f (x) with respect to k, and set k = 1, then we have the
desired result. 2

Theorem: If f is (positively) homogeneous of degree r and differentiable, ∂f (x)/∂xi is homo-

geneous of degree r − 1.
Proof. Clear from the definition of partial derivative. 2

Def. When the production function exhibits Constant Returns to Scale (crs), it is positively
homogeneous of degree one (or has the first degree of homogeneity). It means that if material
inputs, machines, and labour inputs are increased by the same percentage, then the output will
also increase by the same proportion.
Profit Maximization: Profit maximization is compatible with constant returns to scale only
when maximized profit is zero.
Calculation of Profit Maximization:
Production function: F (x1 , x2 , . . . , xn , k1 , k2, . . . , km , `1, `2, . . . , `s );
xi : material inputs, kj : durable machines, `t : labor power
pi : price of material inputs, rj : rental of durable machines, wt : wage rate of labor power of
type t.
p = the price of product P P Ps
Profit π ≡ p·F (x1 , x2, . . . , xn , k1 , k2 , . . . , km , `1 , `2 , . . . , `s )−( ni=1 pi ·xi + m
j=1 rj ·kj + t=1 wt ·`t )
Solve the following system of simultaneous equations:
 ∂π
 ∂xi = 0, i = 1, . . . , n
∂π
= 0, i = 1, . . . , m .
 ∂k∂πi
∂`i
= 0, i = 1, . . . , s

As a typical equation, we have

∂F
p· − pi = 0.
∂xi
This says the value of marginal product of material input i is equal to the price of the input
i: the latter being the marginal cost of the material input.
NB. marginal revenue = marginal cost, at the profit maximizing operation.
Def. Fixed cost is the expense necessary, independent of production level.
NB. The existence of fixed cost does not damage the homogeneity of degree zero of supply
functions. Why?
Def. Variable cost is the expenditure which varies with production level.

Exercises:
Q01. Let the production function be F (x1 , x2) = A · xa1 · xb2 , the price of product p, the cost
function (q1 · x1 + q2 · x2 ). Find out x∗1 and x∗2 which maximize profits. The prices p, q1 ,and
q2 are given, while A, a, and b are constants.. ( a > 0, b > 0, and a + b < 1. Thus, decreasing
returns to scale (drs) obtain.)
Answer. The profits can be expressed as π ≡ p · A · xa1 · xb2− (q1 · x1 + q2 · x2 ). We differentiate
this wrt x1 and x2 , getting
 July 27, 2009

½ ∂π
∂x1
= p · A · a · xa−1
1 · xb2 − q1 = 0 (1)
.
∂π
∂x2
= p · A · b · xa1 · xb−1
2 − q2 = 0 (2)

From eqs.(1) and (2), we have

½
p · A · a · xa1 · xb2 = q1 · x1 (3)
.
p · A · b · xa1 · xb2 = q2 · x2 (4)

Thus, we have the now familiar equation in maximizing a function of Cobb-Douglas type:
a q1 · x1
= .
b q2 · x2
Multiplying eq.(3) by q2b , we obtain p · A · a · xa1 · (q2 · x2)b = q1 · q2b · x1 , form which it follows
p · A · a · xa1 · ( ab · q1 · x1)b = q1 · q2b · x1. Thus,
pA 1
x∗1 = ( · a1−b · bb ) 1−(a+b) .
q11−b b
· q2
Similarly,
pA 1
x∗2 = ( 1−a · aa
· b 1−a 1−(a+b)
) .2
q1a · q2
Q02. Confirm the solutions x∗1 and x∗2 are homogeneous of degree zero wrt prices p, q1 , and q2.
Q03. Confirm the optimum output is also homogeneous of degree zero wrt prices p, q1, and q2.

Cost Functions Approach:

Often, profit is described by the following, using the cost function which depends upon the
quantity produced.

π ≡ p · x − C(x), C(x) ≡ ax3 + bx2 + cx + d, where a > 0, b < 0, c > 0, d ≥ 0, b2 − 3ac ≤ 0.

Here d is the fixed cost. The output x is the sole variable with the parameters p, a, b, c, d given.
Exercises:
Q04. Depict a graph of the cost function above.
Q05. Suppose p = 10, and C(x) = x3 − 5x2 + 10x + 10. Compute the profit maximizing output
and the profit.
Q06. Suppose p = 7, and C(x) = x3 − 5x2 + 10x + 10. Compute the profit maximizing
output and the profit. Should the firm carry out the production?
Q07. Suppose p = 5, and C(x) = x3 − 5x2 + 10x + 10. Compute the profit maximizing
output and the profit. Should the firm carry out the production?
NB. Even if the maximum profit is negative, it is better than doing nothing because of the
fixed cost. This is in the short-run phenomenon, though. Consider how such a situation can
disappear.

“Red” Herring:
Important formula of differentiation: the Quotient Rule:

d( u(x)
v(x)
) u0 · v − u · v 0 du(x) dv(x)
= 2
, where u0 ≡ and v 0 ≡ .
dx v dx dx
 July 27, 2009

Convergence of sequence:

{xn } → x∗ iff for any ² > 0, there exists a N > 0 such that kxn − x∗ k < ².

This can be written, using logics symbols, as

{xn } → x∗ iff ∀ ² , ∃ N such that kxn − x∗ k < ².

Exercises:
Q08. Prove the sequence {2−n } , (n = 1, 2, . . .) converges to 0.
Q09. Consider the sequence {(−1)n · (1 − n1 )}, (n = 1, 2, . . .). Prove that this does not converge
to 1. The subsequence {(−1)k } (k = 2, 4, . . . , 2m, . . .), however, converges to 1.

16 Cost Function
Suppose C(x) = x3 − 5x2 + 10x + 10. Find out the price of product at which the maximized
profit is zero. Thus, the profit maximizing output is ‘break-even’. Find out also the price at
which the firm stops production.
NB. Please produce a report about this problem.
Answer.
Let the cost function be, in general,

C(x) = x3 + bx2 + cx + d. ( b < 0, c > 0, d > 0, and b2 − 3c ≤ 0).

( b < 0 to have the positive inflection point; b2 − 3c ≤ 0 for the cost not to decrease.)
At the output level xo , the tangent line to the cost curve is:

y − (x3o + bx2o + cxo + d) = (3x2o + 2bxo + c) · (x − xo ). (L)

In order for this tangent line to go through the origin, i.e., when x = 0, y = 0, it is necessary to
have

−(x3o + bx2o + cxo + d) = −(3x3o + 2bx2o + cxo ).

That is,

2x3o + bx2o − d = 0. (*)

In the special case where b = −5, c = 10, and d = 10, the equation (*) becomes

2x3o − 5x2o − 10 = 0

This has only one positive solution, while other two are complex numbers. The positive solution
is: xo ∼
= 3. 0408. Then the price of product is:

3(3.04)2 + 2(−5)(3.04) + 10 ∼
= 7.3248.

So, at the price 7.3248 (or about 7.3), the maximized profit is 0 at the output about
3.04.
Next, let us find out the price at which the firm stop its operation. This is when the tangent
line at the profit maximizing output go through the point (0, d). Thus, we substitute 0 for x,
and d for y in the equation (L) above.
 July 27, 2009

d − (x3o + bx2o + cxo + d) = (3x2o + 2bxo + c) · (0 − xo ). (L)

That is,

2x3o + bx2o = 0. (**)

In our particular case,

2x3o − 5x2o = 0.

We solve this equation, and find out xo = 2.5. (Certainly, xo = 0 also, but no operation at any
price lower than 2.5.)
The price of the product at which the firm makes loss as large as the fixed cost, 10, is:

3(2.5)2 + 2(−5)(2.5) + 10 = 3.75.

So, at the price 3.75, the maximized profit is −10 at the output 2.5.2

17 Cost Function: the Case of Quadratic Equation

Let us treat an easier case. Suppose C(x) = ax2 + bx + c, with a > 0, b ≥ 0, and c ≥ 0. Then,
the profit function is:

π(x) ≡ p · x − C(x),

where p is the price of product. In this case, there is no positive price of the product, above
which the production is carried out even making losses in order to avoid the full expenses on
fixed cost.
In general, the profit is maximized at
dπ p−b
= −2ax + (p − b) = 0, i.e., x = .
dx 2a
When the cost function is given as above, the tangent line at the output level, xo ,is

y − (ax2o + bxo + c) = (2axo + b)(x − xo ).

This tangent line go through the origin if y = 0 at x = 0. This means

−(ax2o + bxo + c) = −2ax2o − bxo , i.e., ax2o − c = 0.

Thus,
r
c
xo = .
a
The slope at this point is
r
c √
2a + b = 2 ac + b.
a
 July 27, 2009

Therefore, given the cost function C(x) = ax2 + bx + c, if the price satisfies
√
p = 2 ac + b,
p
then the maximum profit is zero at the output xo = ac .

Exercises:
Q01. Suppose the cost function is C(x) = x2 + 2x + 5. The price of product is p = 10̇. Find
out the profit maximizing output and the profit there.
Q02. Suppose the cost function is C(x) = x2 + 2x + 7. Find out the price of product at which
the maximized profit becomes zero.

18 Cubic Equation
In the calculation of zero-profit price, we have come to the following cubic equation:

2x3 + bx2 − d = 0.

A real positive solution for this is :

 

 sµ ¶ 

 1 1 3 1p 1 b 2
1 
x= 3
d− b + (81d2 − 3db3 ) + r ´ − 6 b .

 4 216 36 36 3 ³ 1 p 
 d− 1 3 1 2 3
b + 36 (81d − 3db ) 
4 216

In general, a real solution to

x3 + bx2 + d = 0,

is:
 

 sµ ¶ 

 1 1 3 1p 1 b2 1 
x= 3
− d− b + (81d + 12db ) + r³
2 3
´ − 3 b ,

 2 27 18 9 3 p 
 − 12 d − 1 3 1 2 3
b + 18 (81d + 12db ) 
27

while a real solution to

x3 + cx + d = 0,

is:
 

 sµ ¶ 

 1 1p 1 c 
x= 3
− d+ (12c3 + 81d2) − r³ ´ .

 2 18 3 3 p
 − 12 d + 1
18
(12c + 81d ) 
3 2 
 July 27, 2009

19 Newton-Raphson Method
To solve linear or nonlinear equations, we can use the Newton-Raphson method under some
conditions. This is an iterative method, and we choose an initial value near the wanted solution.
We draw a tangent line at (x(0) , f(x(0) )) , and the point where this line cut the x-axis is the
next point. This process is repeated until a point gets near the wanted solution.
The method is described by:

f(x(i) )
x(i+1) = x(i) − , (1)
f 0 (x(i))
df
where f 0 = dx
.

NB. The Newton-Raphson method may not converge. You need some conditions to secure the
convergence.

The method was generalized to the n variable case by Leonid Kantrovich (Nobel Prize in
1975):

x(i+1) = x(i) − (JF (x(j) ))−1 · F (x(i) ),

where JF (x(j) ) is the Jacobian of F with respect to x evaluated at x(i) .

20 Gale-Nikaido’s Proof of Existence of General Equilib-

rium
We prove the following system of equations has a solution, using the Brouwer Fixed Point
Theorem.


 e1 (p1 , p2, . . . , pn ) ≤ 0,

e2 (p1 , p2, . . . , pn ) ≤ 0,
(GE)

 ···

en (p1 , p2 , . . . , pn ) ≤ 0,

where ei (p) ≡ di (p) − si (p) is the excess demand function for commodity i, with p ≡
(p1 , p2 , . . . , pn ). We know ei ’s are continuous on the non-negative orthant of Rn , and homo-
geneous of degree zero. We also have the Walras Law:
X
n
pi · ei (p) = 0.
i=1

Therefore, a solution to the system (GE), p, satisfies the following:

(1) if pi > 0, then ei (p) = 0, and (2) if ej (p) < 0, then pj = 0.

We define the (n − 1)-simplex S to be

X
n
S ≡ {x | x ∈ Rn+ , xi = 1}.
i=1
 July 27, 2009

We construct the following map T from S into itself:

pi + max(0, ei (p))
T :p∈S →( P ).
1 + ni=1 max(0, ei (p))

This map T is continuous, and the set S is compact convex, and so there is at least one fixed
point p∗ . This should be an equilibrium price vector. First we show at the fixed point p∗ , the
denominator is 1. Suppose the denominator is k, and k > 1. (It is evident k ≥ 1.) Then we
get

(k − 1) · pi = max(0, ei (p)) for all i.

This means that if pi > 0, then ei (p) > 0, which is a contradiction to the Walras Law.2

Reference:
Uzawa, H. (1962): “Walras’ Existence Theorem and Brouwer’s Fixed Point Theory”, Economic
Studies Quarterly, vol. 31.
Exercise:
Q01. Try to devise out your own map whose fixed point can be an equilibrium.

21 Supplement: Excess Demand Functions–Examples

Reference:
T. Fujimoto, “Numerical Examples of Systems of Excess Demand Functions”, Kagawa University
Economic Review, vol.60, 1988, pp.801-805.

Assumption: We consider three commodity economies only.

˙ In
The equilibrium price vector is common among the first 4 examples and it is P ∗ = (1, 1, 1).
each case, verify the homogeneity of degree zero and the Walras Law.

(1) Gross Substitutes


1/P1 P2 /P12 P3 /P12

 E1 (P) = ( A 2 − 1) + A + A 2 ,
P /P P /P
E2 (P) = 1A 2 + ( 1/P 2
− 1) + 3A 2 ,

 P /P 2
A
P /P 2
E3 (P) = 1A 3 + 2A 3 + ( 1/P A
3
− 1),

where A ≡ 1/P1 + 1/P2 + 1/P3 .

The sign pattern of the Jacobian matrix:
 
− + +
 + − + .
+ + −

NB. When there are three consumers, and the common utility function of individual is u(x1, x2 , x3 ) =
√ √ √
x1 + x2 + x3 . with the initial endowment of individual i being one unit of commodity i.

(2) Negative Dominant Diagonal

 July 27, 2009


P22 P32

 E1 (P) = 2 + 2 − 1,
P1 +P1 P2 P1 +P1 P3
P1 P2
E2(P) = − ,

 E (P) = P2 +P3
P1
P1 +P2
P3
3 P2 +P3
− P1 +P3
.

The sign pattern of the Jacobian matrix:

 
− + +
 + − − .
+ − −

NB. This system also satisfies the weak axiom of revealed preference.

(3) Giffen Goods (NB. Unbounded from below)


P2 P2

 E1(P) = 1 − P12 +P2 1 P2 − P12 +P3 1 P3 ,
E2 (P) = P1P+P
2
− P2P+P1
,

 E (P) = P3 − P1 3 .
2

3 P1 +P3 P2 +P3

The sign pattern of the Jacobian matrix:

 
+ − −
 − + + .
− + +

Cf. inferior goods.

(4) Scarf’s Example (NB. Global Instability)

H. Scarf, “Some Examples of Global Instability of Competitive Equilibrium”, International
Economic Review, vol.1, 1960, pp.157-172.
 P3
 E1 (P) = P3 +P 1
− P1P+P
2
2
,
P1 P3
E2 (P) = P1 +P2 − P2 +P3 ,

E3 (P) = P2P+P
2
3
− P3P+P
1
1
.

The sign pattern of the Jacobian matrix:

 
± − +
 + ± − .
− + ±

(5) Nikaido’s Example (Gross substitutes)

H. Nikaido, Convex Structures and Economic Analysis, Academic Press, New York, 1968.
X
n X
n
Ei (P) = aij Pj /Pi − aji ., for a given n × n positive matrix A ≡ (aij ).
j=1 j=1

When A is given as
 July 27, 2009

 
1 2 3
A =  4 3 2 ,
5 5 5


 E1 (P) = (P1 + 2 · P2 + 3 · P3 )/P1 − 10,
E2 (P) = (4 · P1 + 3 · P2 + 2 · P3)/P2 − 10,

E3 (P) = (5 · P1 + 5 · P2 + 5 · P3)/P3 − 10.

NB. The equilibrium price vector is P ∗ = (5, 6, 11).

Exercises:
Q01. Verify the weak axiom of revealed preference for the system in (2) above.
Q02. Compute the equilibrium price vector in your own way for the example in (4) Nikaido’s
example.

22 Input-Output Models: A Simple Leontief Model

References:
Leontief, Wassily: The Structure of the American Economy, 1919-1939, Oxford Univ Press, 1941.
Dorfman, Robert, Paul A. Samuelson and Robert M. Solow: Linear Programming and Economic
Analysis, McGraw-Hill, 1958.
Nikaido, Hukukane, Linear Algebra for Economics (in Japanese “Keizai no tameno Senkei
Suugaku”), Baifukan, 1961.

Assumptions:
(1) the absence of joint production; (2) the absence of durable capital goods;
(3) the homogeneous labor; (4) constant returns to scale;
(5) the absence of externalities; (6) the uniform production period called year;

Variables and Data:

x : an n-column vector of activity levels, a gross output vector;

p : an n-row vector of wage-unit prices;

the given n by n input-output matrix:
A :  aij = the amount of commodity i necessary to produce
one unit of commodity j in industry j.
` : the given n-row vecor of labor input coefficients;
g : a given rate of steady balanced growth;
r : a given rate of uniform profit;
d : a given n-column vector of final demands.

Quantity Equation: the gross output of each commodity at the end of production period
equals the input at the beginning of the next period plus the final demand.

x = (1 + g) · Ax + d.
 July 27, 2009

Price Equation: the price of a commodity equals its cost, that is, the cost of material inputs
with a mark-up ratio, r, plus the labor cost.

p = (1 + r) · pA + `.

23 A Two Commodity Model

We consider a special case of two commodities, say, corn and tractor.

x = (x1 , x2 )0 ,
p = (p1 , p2 ),
µ ¶
a11 a12
A = ,
a21 a22
` = (`1 , `2 ),
d = (d1 , d2)0 , and
g = r = 0.

Then, the quantity equation, x = Ax + d, becomes

½
x1 = a11 · x1 + a12 · x2 + d1 ,
x2 = a21 · x1 + a22 · x2 + d2.

And the price equation, p = pA + `,

½
p1 = a11 · p1 + a21 · p2 + `1 ,
p2 = a12 · p1 + a22 · p2 + `2.

NB. The columns of A show the inputs of industries.

Exercises:
Q01. Let µ ¶ µ ¶
0.2 0.5 1
A≡ , ` ≡ (1, 1), and d ≡ .
0.3 0.6 0
(1) Solve the quantity and price equations with g = r = 0.
(2) When d becomes d = (0, 1)0 , what are solutions to the quantity equation?

Q02. Let A, `, and d be as in Q01. Does the quantity equation have a meaningful solution
when g = 0.1? How about when g = 0.3? What is the maximum rate of growth below which
the quantity equation has a semi-positive solution?
Q03. Compute the inverse of the matrix (I − A).

24 Input-Output Models: How to Solve It

(1) In general:
Let us consider the system of simultaneous equation M x = d , where
 July 27, 2009

     
m11 m12 · · · m1n x1 d1
 m21 m22 · · · m2n   x2   d2 
     
M ≡ ... , x =  ..  , and d =  .. .
 ··· ··· ···   .   . 
mn1 mn2 · · · mnn xn dn

The matrix M and the vector d are given, while x is a vector of unknowns.
This system can be solved by use of the inverse of M , if M is regular, i.e., |M| 6= 0. That
is,

x = M −1 · d.

You are reminded of the solution x = m−1 · d for a scalar equation mx = d. Another formula
to solve the equation is Cramer’s Rule:
i
¯ ∨ ¯
¯ .. . ¯
¯ . d .. ¯
(Cramer’s Rule) xi = ,
|M|
where the numerator is M with the i-th column replaced by d.
Determinants:
A given matrix is 1 × 1, m, then its determinant is m. Then, for a n × n matrix, its determinant
is defined recursively as
X
n
|M| ≡ aij · (−1)i+ j · |Mij | ,
j=1

where Mij is the (n − 1) × (n − 1) matrix created by deleting the i-th row and the j-th column
from M . This is the expansion of |M |, using the i-th row. The expansion via j-th column is
X
n
|M| ≡ aij · (−1)i+ j · |Mij | .
i=1

Normally, we use a column or a row which has many zeros as possible.

Calculation of the Inverse using the Cramer’s Rule:

When d is
 . 
..
 
 1 
d = ei = 
 ...
,

 
..
.

i.e., d has one in the i-th element with the remaining elements being all zero, the solution to
Mx = d is the i-th column of the inverse M −1 .

(2) 2 by 2 case:
Let
 July 27, 2009

µ ¶
a b
M≡ ,
c d
then
à !
d −b
M −1 = |M |
−c
|M |
a .
|M | |M |

The determinant |M | is

|M | = ad − bc.
Cranes-and-Tortoises Calculation (“Tsuru-Kame Zan”):
½
Cranes + Tortoises = 20,
(Legs of Cranes) + (Legs of Tortoises) = 62.
Suppose there are 20 cranes, then we have 40 legs. By replacing one crane by one tortoise, we
can increase two legs, and 22(=62-40) legs should be added, and so the number of tortoises is
11(=22/2). And that of cranes is 9.
The above can be converted to a system of simultaneous equations:
x =number of cranes, and y =number of tortoises.
½ ½µ ¶µ ¶ µ ¶
x + y = 20, 1, 1 x 20
, i.e., = .
2x + 4y = 62. 2, 4 y 62
µ ¶ µ ¶ µ ¶
1, 1 −1 4/2, −1/2 2, −1/2
M= , |M | = 1 × 4 − 1 × 2 = 2, and M = = .
2, 4 −2/2, 1/2 −1, 1/2
Thus,
µ ¶ µ ¶ µ ¶ µ ¶
x −1 20 40 − 31 9
=M · = = .
y 62 −20 + 31 11
Or, using the Cramer’s Rule:
¯ ¯ ¯ ¯
¯ 20, 1 ¯ ¯ 1, 20 ¯
¯ ¯ ¯ ¯
¯ 62, 4 ¯ 80 − 62 ¯ 2, 62 ¯ 62 − 40
x= = = 9, y = = = 11.
|M | 2 |M| 2

(3) the case of (I-A): We know A ≥ 0. If there exists an x ∈ R+n such that x À Ax. Then,

(I − A)−1 = I + A + A2 + · · · + Ak + · · · (Neumann series).

For the quantity equation, x = Ax + d, we have an iterative method:

x(0) = d,
x(1) = Ax(0) + d,
···
(k+1)
x = Ax(k) + d,
··· .
 July 27, 2009

Exercises:
Q01. Find out the numbers of cranes and tortoises which satisfy the following:
½
Cranes + Tortoises = 30,
(Legs of Cranes) + (Legs of Tortoises) = 102.

25 Answers to the Problems:

1. Consider the case of three commodities, and u(x, y, z) = 2x + 3y + 6z, px = 20, py = 60, pz =
100, and E = 12000. Derive the utility maximizing consumption basket. Now suppose the
price of x starts to rise, and you wish to find out the price of x above which commodity x is
substituted for by another commodity. By which commodity, is x substituted?
Hint. Lecture note (3). Draw a figure of indifference surface.
Answer:
If we buy only commodity x, we can buy qx = 12000 20
= 600, giving the utility level, ux:only =
2 × 600 = 1200. Next, in a similar way, with commodity y, we can buy qy = 12000 60
= 200, giving
the utility level, uy:only = 3×200 = 600. Third, with commodity z, we can buy qz = 12000100
= 120,
giving the utility level, uy:only = 6 × 120 = 720.
So, the utility maximizing basket is: (600, 0, 0), purchasing commodity x only. The second
highest utility is realized by buying commodity y only. Thus, when the price of commodity x
rises up to a point where it gives the utility level 720, we become indifferent between x and z.
This level is obtained by getting the amount 720 2
= 360 of x. This amount is purchasable when
its price is 12000
360
= 33. 333.
The utility maximizing basket is: (600, 0, 0). At px = 34, commodity x is substituted
by z.

2. Consider the case of three commodities and u(x) = xα1 · xβ2 · xγ3 , where α > 0, β > 0, and
γ > 0, and derive the utility maximizing consumption basket subject to the budget constraint ,
p1 · x1 + p2 · x2 + p3 · x3 = E, where p1 > 0, p2 > 0, p3 > 0 and E > 0.
Hint. Use the Lagrangian multiplier method.
Answer. Rewrite the constraint into the form p1 · x1 + p2 · x2 + p3 · x3 − E = 0, and define the
Lagrangian functions as

L ≡ xα1 · xβ2 · xγ3 − λ · (p1 · x1 + p2 · x2 + p3 · x3 − E).

Differentiate L wrt each variable xi and the Lagrangian multiplier λ, and set them to zero, thus
obtaining a system of four equations with four variables.



∂L
= α · xα−1 · xβ2 · xγ3 − λ · p1 = 0, (1)

 ∂x1 1
∂L
∂x2
= β · xα1 · xβ−1
2 · xγ3 − λ · p2 = 0,



∂L
∂x3
= γ · xα1 · xβ2 · xγ−1
3 − λ · p3 = 0,
 ∂L
∂λ
= −( p1 · x 1 + p 2 · x 2 + p3 · x3 − E) = 0.

Multiply eq.(1) by x1 , then we get α · xα1 · xβ2 · xγ3 − λ · p1 · x1 = 0. (NB.Revive the original utility
function.) Thus, it follows

α · xα1 · xβ2 · xγ3

p 1 · x1 = .
λ
In a similar way, we have
 July 27, 2009

β · xα1 · xβ2 · xγ3 γ · xα1 · xβ2 · xγ3

p2 · x2 = and p3 · x3 = .
λ λ
Hence

(p1 · x1 ) : (p2 · x2 ) : (p3 · x3 ) = α : β : γ .

Ans so, finally, we have

1 α 1 β 1 γ
x1 = · · E, x2 = · · E and x3 = · · E .2
p1 α + β + γ p2 α + β + γ p3 α + β + γ
PS. (1) The price rise in a commodity has no effect on the consumption of other commodities.
(2) When the expenditure E increases, the consumption of every commodity goes up.
Q
2∗ . Consider the case of 100 commodities and u(x) = 100 i 1 2 3 100
i=1 xi = x1 · x2 · x3 · · · · · x100 , and derive
the utility maximizing consumption basket subject to the budget constraint , p1 · x1 + p2 · x2 +
p3 · x3 + . . . + p100 · x100 = E, where p1 = 1, p2 = 2, p3 = 3, . . . , p100 = 100, and E = 5050.
t t
Answer. The expenditure to t-th commodity, pt · xt = P 100 · E = 5050 · 5050. Therefore, xt = 1
i=1 i
because pt = t. 2P
NB. A formula: ni=1 i = n(n+1) 2
.

3. Prove that a monotone non-decreasing map from the real interval [0,1] into itself has at least
one fixed point. Hint. Lecture note (4).
Answer to 3: Omitted. (Consult your own notebook.)

4. Let the production function be F (x1 , x2 ) = A · xα1 · xβ2 , the price of product be p, the cost
function C(x1, x2 ) = q1 · x1 + q2 · x2. Find out the output x = (x1, x2 ) which maximize profits.
All the given constants are positive, i.e., A > 0, α > 0, β > 0, q1 > 0, q2 > 0, and α + β < 1.
(Thus, decreasing returns to scale (drs) obtain.)
Hint. Lecture note (6).

5. Suppose the cost function is C(x) = x3 − 5x2 + 10x + 12. Find out the price of product at
which the maximized profit is zero. Thus, the profit maximizing output is ‘break-even’.
Find out also the price at which the firm stops production. Hint. Lecture note (7).

Answer to 5:
Let the cost function be, in general,

C(x) = x3 + bx2 + cx + d. ( b < 0, c > 0, d > 0, and b2 − 3c ≤ 0).

At the output level xo , the tangent line to the cost curve is:

y − (x3o + bx2o + cxo + d) = (3x2o + 2bxo + c) · (x − xo ). (L)

In order for this tangent line to go through the origin, i.e., when x = 0, y = 0, it is necessary to
have

−(x3o + bx2o + cxo + d) = −(3x3o + 2bx2o + cxo ).

 July 27, 2009

That is,

2x3o + bx2o − d = 0. (*)

In the special case where b = −5, c = 10, and d = 12, the equation (*) becomes

2x3o − 5x2o − 12 = 0

This has only one positive solution, while other two are complex numbers. The positive solution
is: xo ∼
= 3. 1174. Then the price of product is:

3(3.12)2 + 2(−5)(3.12) + 10 ∼
= 8.0032.

So, at the price 8.0032 (or about 8.0), the maximized profit is 0 at the output about
3.12.
Next, let us find out the price at which the firm stop its operation. This is when the tangent
line at the profit maximizing output go through the point (0, d) [An error was here]. Thus, we
substitute 0 for x, and d for y in the equation (L) above.

d − (x3o + bx2o + cxo + d) = (3x2o + 2bxo + c) · (0 − xo ). (L)

That is,

2x3o + bx2o = 0. (**)

In our particular case,

2x3o − 5x2o = 0.

We solve this equation, and find out xo = 2.5. (Certainly, xo = 0 also, but no operation at any
price lower than 2.5.)
The price of the product at which the firm makes loss as large as the fixed cost, 12, is:

3(2.5)2 + 2(−5)(2.5) + 10 = 3.75.

So, at the price 3.75, the maximized profit is −12 at the output 2.5.2
verification: 3.75 × 2.5 − ((2.5)3 − 5 × (2.5)2 + 10 × 2.5 + 12) = −12.0

NB. In this problem, we changed the fixed cost, i.e., d, only. Please consider the effects on the
output, and the ‘break-even’ price of the increase in the fixed cost. Hint. Draw a diagram.

6. Explain the Newton-Raphson method to solve nonlinear equations.

Hint Lecture notes (9) and (14)*.
Answer: This is an iterative method to solve a nonlinear equation of a single variable, f (x).
We choose an initial value, x(0) , close to the desired solution. Then, we consider the tangent line
at (x(0), f(x(0) )), and the cross point between this tangent line and the x-axis is made the next
iterate x(1). We continue this procedure until a satisfactory approximation is attained.
NB. A generalization to many variables was made by Leonid Vitaliyevich Kantrovich (1912-1986;
Nobel Prize, 1975).
Newton-Raphson Method:
2x3o − 5x2o − 12 = 0.
 July 27, 2009

formula:
f(x(i) )
x(i+1) = x(i) − .
f 0 (x(i))

2 × (x(i) )3 − 5 × (x(i))2 − 12
x(i+1) = x(i) − .
6 × (x(i) )2 − 10 × (x(i) )

We start from x(0) = 4. Then,

2 × (4)3 − 5 × (4)2 − 12
x(1) = 4 − ' 3.3571.
6 × (4)2 − 10 × (4)

2 × (3.36)3 − 5 × (3.36)2 − 12
x(2) = 3.36 − ' 3.1427.
6 × (3.36)2 − 10 × (3.36)

2 × (3.14)3 − 5 × (3.14)2 − 12
x(3) = 3.14 − ' 3.1177.
6 × (3.14)2 − 10 × (3.14)

Continue the procedure until the tolerance limit you have designated.

References:
Isaac Newton (1643-1727): Methodus Fluxionum et Serierum Infinitarum, (1664-1671).
Joseph Raphson (1648-1715): Analysis Aequationum Universalis, London, 1690.
A. S. Householder, The Numerical Treatment of a Single Nonlinear Equation, McGraw-Hill:
New York, 1970.
L.V. Kantorovich and G.P. Akilov, Functional Analysis in Normed Spaces, Pergamon Press:
New York, 1964.
J.M. Ortega and W.C. Rheinboldt, Iterative Solution of Non Linear Equations in Several
Variables, Academic Press: New York, 1970.
H. Qiu, “A Robust Examination of the Newton-Raphson Method with Strong Global
Convergence Properties”, Master’s Thesis: University of Central Florida, 1993.
Tjalling J. Ypma, “Historical Development of the Newton-Raphson Method”, SIAM Review,
vol. 37(4), pp. 531-551, 1995.
¥
Exam: 2009-8-21(Fri).