Scribd
Upload
137 views18 pages

07 Convex Analysis

This document provides an overview of concepts in convex analysis including: 1) Definitions of convex sets and convex combinations 2) Properties of continuous and differentiable functions on convex sets 3) Characterizations of concave functions in terms of inequalities and Hessian matrices 4) Relationships between concave functions and convex sets 5) Definitions and properties of quasi-concave functions The document includes theorems, examples, and exercises related to these topics.

Uploaded by

alin444444
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
137 views18 pages

07 Convex Analysis

This document provides an overview of concepts in convex analysis including: 1) Definitions of convex sets and convex combinations 2) Properties of continuous and differentiable functions on convex sets 3) Characterizations of concave functions in terms of inequalities and Hessian matrices 4) Relationships between concave functions and convex sets 5) Definitions and properties of quasi-concave functions The document includes theorems, examples, and exercises related to these topics.

Uploaded by

alin444444
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 18

Real Analysis:

Convex Analysis

1. Convex Sets
Line Segment: If x; y 2 <n; the line segment joining x and y is given by the set of
points
fz 2 <n : z = x + (1
) y; for some 0
1g :

Convex Set: A set S


<n is a convex set if for every x; y 2 S; the line segment
joining x and y is contained in S:
For example, the set of points (x; y) 2 <2: x2 + y 2

1 is a convex set.

But the set of points (x; y) 2 <2: x2 + y 2 = 1 is not a convex set.


#1. Examples: Suppose two sets S1 and S2 are convex in <n:
(a) Prove that the intersection of S1 and S2, that is, the set
is a convex set in <n:

S = fz 2 <n: z 2 S1 and z 2 S2g ;

(b) Prove that the sum of S1 and S2, that is, the set

S = fz 2 <n: z = x + y; where x 2 S1 and y 2 S2g ;

is a convex set in <n:

(c) Prove that the Cartesian product of S1 and S2, that is, the set

S = z 2 <2n: z = (x; y) ; where x 2 S1 and y 2 S2 ;

is a convex set in <n:

(d) Is the union of S1 and S2, that is, the set


a convex set in <n?

S = fz 2 <n: z 2 S1 or z 2 S2g ;

Convex Combination: A vector y 2 <n is said to be a convex combination of the


vectors x1; x2; :::; xm 2 <n if there exist m non-negative real numbers 1; 2; ::: m such
that
m
m
X
X
i
(i)
i = 1; and (ii) y =
ix :
i=1

i=1

A convex set A <n can then be rede ned as a set such that for every two vectors
in the set A; all convex combinations of these two vectors are also in the set A:
It can be shown that in the above statement two can be replaced by m where m
is any integer exceeding 1:

Proposition 1:
A set S <n is convex if and only if for any integer m > 1 and for any m vectors in
S; every convex combination of these m vectors is in S:
Proof: To be discussed in class.
Hints: Proof by induction.
- Step 1: Show that the property is true for m = 2:
- Step 2: Assume that the property is true for m = k; then show that it is also true
for m = k + 1:

2. Continuous & Differentiable Functions on Convex Sets


We now provide three very useful theorems on continuous and differential functions
on convex sets.
Theorem 1 (Intermediate Value Theorem):
Suppose A is a convex subset of <n; and f : A ! < is a continuous function on A:
Suppose x1 and x2 are in A; and f x1 > f x2 : Then, given any c 2 < such that
f x1 > c > f x2 ; there is 0 < < 1 such that f x1 + (1
) x2 = c:
#2. Example: Suppose X = [a; b] is a closed interval in < (with a < b); and f : X !
< is a continuous function. Use Weierstrass Theorem and the Intermediate Value
Theorem to prove that f (X) is itself a closed interval.
Theorem 2 (Mean Value Theorem):
Suppose A is an open convex subset of <n; and f : A ! < is continuously differentiable on A: Suppose x1 and x2 are in A: Then there is 0
1 such that

f x2

f x1 = x2

x1

Of

x1 + (1

) x2 :

#3. Example: We say that f : A ! < (where A


ever x1 > x2; we have f x1 > f x2 :

<) is an increasing function if when-

(a) Suppose that g : A ! <, A


<; A is open and convex, and g 0 (x) > 0; for all
x 2 A: Prove, using the Mean Value Theorem, that g is an increasing function.
(b) Suppose g is an increasing function on A: Does it follow that g 0 (x) > 0; for all
x 2 A?
Theorem 3 (Taylor's Expansion upto Second Order):
Suppose A is an open convex subset of <n; and f : A ! < is twice continuously
differentiable on A: Suppose x1 and x2 are in A: Then there is 0
1 such that

f x

f x

= x

Of x

1 2
+ x
2

x1

Hf

x1 + (1

) x2

x2

x1 :

3. Concave Functions
Let A be a convex set in <n: Then f : A ! < is a concave function (on A) if for all
x1; x2 2 A; and for all 0
1;

x1 + (1

) x2

f x1 + (1

) f x2 :

The function f is strictly concave on A if

x1 + (1

) x2 > f x1 + (1

whenever x1; x2 2 A; x1 6= x2; and 0 <

) f x2

< 1:

Let A be a convex set in <n: Then f : A ! < is a convex function (on A) if for all
x1; x2 2 A; and for all 0
1;

x1 + (1

) x2

f x1 + (1

) f x2 :

The function f is strictly convex on A if

x1 + (1

) x2 < f x1 + (1

whenever x1; x2 2 A; x1 6= x2; and 0 <

< 1:

) f x2

Relation between Concave Functions and Convex Sets:


The relation between concave function and convex sets is given by the following
results.
Theorem 4:
Suppose A is a convex set in <n and f : A ! <: Then f is a concave function if and
only if the set {(x; ) 2 A < : f (x)
g is a convex set in <n+1:
Proof: To be discussed in class.
Theorem 5: Suppose A is a convex set in <n and f : A ! < is a concave function.
Then, for every 2 <; the set S ( ) = fx 2 A : f (x)
g is a convex set in <n:
Proof: To be discussed in class.

A result on concave functions which parallels Proposition 1 on convex sets can now
be noted. It is known as Jensen's Inequality, and is a very useful tool in convex
analysis.
Proposition 2 (Jensen's Inequality):
Suppose A is a convex set in <n and f : A ! < is a concave function. Then, for any
integer m > 1,
m
m
P
P
i
i
f
ix
if x
i=1

whenever x ; x ; :::; x 2 A; ( 1;
1

2 ; :::; m )

i=1

<m
+

and

Proof: To be discussed in class.

m
P

= 1:

i=1

Hints: Proof by induction.


- Step 1: Show that the property is true for m = 2:
- Step 2: Assume that the property is true for m = k; then show that it is also true
for m = k + 1:

10

Continuity of Concave Functions:


In general, if A is a convex set in <n and f : A ! < is concave on A; then f need
not be continuous on A:
#4. Give an example to illustrate the above statement.
But if f is an open convex set in <n and f : A ! < is concave on A; then one can
show that f is continuous on A.
Theorem 6:
Suppose A is an open convex set in <n and f : A ! < is a concave function on A:
Then f is a continuous function on A:
Proof: Homework.

11

Differentiable Concave Functions:


In general, if A is an open convex set in <n and f : A ! < is concave on A; then f
need not be differentiable on A:
#5. Give an example to illustrate the above statement.
If f is continuously differentiable on A; then a convenient characterization for f to
be concave on A can be given in terms of the gradient vector of f:
Theorem 7:
Suppose A <n is an open convex set, and f : A ! < is continuously differentiable
on A: Then f is concave on A if and only if for all x1 and x2 in A

f x2

f x1

Of x1

x2

x1 :

Proof: To be discussed in class.


Corollary 1: Suppose A <n is an open convex set, and f : A ! < is continuously
differentiable on A: Then f is concave on A if and only if for all x1 and x2 in A

Of x2

Of x1

x2

x1

0:

12

It is interesting to note that a characterization of strictly concave functions can be


given by replacing the weak inequalities in Theorem 7 and Corollary 1 with strict
inequalities (for x1, x2 in A with x1 6= x2).
Theorem 8:
Suppose A <n is an open convex set, and f : A ! < is continuously differentiable
on A: Then f is strictly concave on A if and only if for all x1, x2 in A with x1 6= x2;

f x2

f x1 < Of x1

x2

x1 :

Corollary 2:
Suppose A <n is an open convex set, and f : A ! < is continuously differentiable
on A: Then f is strictly concave on A if and only if for all x1, x2 in A with x1 6= x2;

Of x2

Of x1

x2

x1 < 0:

13

Twice Differentiable Concave Functions:


If A <n is an open convex set, and f : A ! < is twice continuously differentiable
on A; then we can nd a convenient characterization for f to be concave on A in
terms of the negative semi-de niteness of the Hessian matrix of f:
Theorem 9:
Suppose A
<n is an open convex set, and f : A ! < is twice continuously
differentiable on A: Then f is concave on A if and only if the Hessian matrix Hf (x)
is negative semi-de nite for all x in A:
Theorem 10:
Suppose A
<n is an open convex set, and f : A ! < is twice continuously
differentiable on A: If the Hessian matrix Hf (x) is negative de nite for all x in A;
then f is strictly concave on A:
#6. Give a counter-example to establish that the converse of Theorem 10 does not
hold.

14

4. Quasi-Concave Functions
Let A
<n be a convex set, and f a real-valued function on A: Then f is quasiconcave on A if
f x2
f x1 implies f x1 + (1
) x2
f x1
whenever x1; x2 2 A; and 0

1:

The function f is strictly quasi-concave on A if

f x2

f x1 implies f

whenever x1; x2 2 A; x1 6= x2; and 0 <

x1 + (1

) x2 > f x1

< 1:

While the conditions stated in Theorem 5 did not characterize concave functions, it
does characterize quasi-concave functions.
Theorem 11:
Suppose A is a convex set in <n and f : A ! <: Then f is a quasi-concave function
on A if and only if for every 2 <; the set S ( ) = fx 2 A : f (x)
g is a convex
set in <n:

15

Differentiable Quasi-Concave Functions:


A characterization of differentiable quasi-concave function can be given which parallels the characterization of differentiable concave functions given in Theorem 8.
Theorem 12:
Suppose A <n is an open convex set, and f : A ! < is continuously differentiable
on A: Then f is quasi-concave on A if and only if for all x1 and x2 in A

f x2

f x1 implies x2

x1

Of x1

0:

Proof: To be discussed in class.


Twice Differentiable Quasi-Concave Functions:
An interesting characterization of twice differentiable quasi-concave functions can be
given in terms of the bordered Hessian matrix associated with the functions.

16

Let A <n is an open convex set, and f : A ! < is twice continuously differentiable
on A: The bordered Hessian matrix of f at x 2 A is denoted by Gf (x) and is de ned
by the following (n + 1) (n + 1) matrix:
1
0
0
Of (x)
A:
Gf (x) = @
Of (x) Hf (x)

We denote the (k + 1)th leading principal minor of Gf (x) by jGf (x; k)j ; where k =
1; 2; :::; n:
Theorem 13:
Suppose A
<n is an open convex set, and f : A ! < is twice continuously
differentiable on A:
(i) If f is quasi-concave on A; then ( 1)k jGf (x; k)j

0 for x 2 A; and k = 1; 2; :::; n:

(ii) If ( 1)k jGf (x; k)j > 0 for x 2 A; and k = 1; 2; :::; n; then f is strictly quasi-concave
on A:

17

References
Must read the following chapters and sections from the textbook:
Section 30.1 (pages 822 827): Weierstrass Theorem and Mean Value Theorems,
Section 30.2 (pages 827 832): Taylor Polynomials on <1;

Section 30.3 (pages 832 836): Taylor Polynomials on <n;

Section 21.1 (pages 505 516): Concave and Convex Functions,


Section 21.2 (pages 516 520): Properties of Concave Functions,
Section 21.3 (pages 522 527): Quasi-concave and Quasi-convex Functions.
This material is based on
1. Nikaido, H., Convex Structures and Economic Theory, (chapter 1),
2. Takayama, A., Mathematical Economics, (chapters 0, 1),
3. Apostol, T., Mathematical Analysis, (chapters 4, 6),
4. Bartle, R., The Elements of Real Analysis, (chapter 7),
5. Mangasarian, O. L., Non-Linear Programming, (chapters 2, 3, 4, 6, 9).

You might also like