Chapter The purpose of this chapter is to

introduce the fundamental concept

Three of convex sets, to describe some
properties of these sets, and to
derive the basic separation theo-
Convex rems for convex sets. These sepa-
ration theorems are the foundations
Sets on which many optimality condi-
inR n tions of nonlinear programming
rest.

1. Convex sets and

their properties
In order to define the concept of a
convex set, we begin by defining
line and line segments through two
points in Rn.
Line
Let x\x2 e R*. The line
through xl and x* is defined as the
set
{x\x = (1 - X)*1 + Xz2, X ER]
or equivalently
[x | x = pix1 + pzx2, pi, pz G R,
Pi + Pz = 1}
If we rewrite the first defini-
tion in the equivalent form
[ x \ x = x1 -f X(z 2 - x1), X e # )
and consider the case when x G R*,
it becomes obvious that the vector
equation x = x1 + X(z 2 — x1) is
the parametric equation of elemen-
tary analytic geometry of the line
through x1 and x2, Fig. 8.1.1.
Convex Seta in 72" 8.1

Line segments
Let xl, x2 £ Rn. We define the following line segments joining z1
z
and x :

(i) Closed line segment [xl,x*] = {x \ x = (1 - X)*1 + \x*, 0 g X ^ 1}

(ii) Open line segment (xl,x2) = {x \ x = (1 - X)z l -f Xz 2 , 0 < X < 1}
(iii) Closed-open line segment [xl,x*) = {x \ x — (1 — X)^1 + Xz2,
0 ^ X < 1}
(iv) Open-closed line segment (xl,x*] = {x \ x = (1 — \)xl + Xz2,
0 < X ^ 1}

Obviously [z^z2] is the portion of the straight line through a;1 and
x which lies between and includes the points x1 and x2, Fig. 3.1.1. (z*,z2)
z

does not include x1 or z2, (a:1,*;2) does not include a:2, and (xl,x*] does not
include x1.
Convex set
A set F C Rn is a convex set if the closed line segmentf joining
every two points of F is in T. Equivalently we have that a set F C Rn
is convex if

Figure 3.1.2 depicts some con vex sets in R2, and Fig. 3.1.3 some non-
convex sets in Rz. It follows from 3 that Rn itself is convex, that the
empty set is convex, and that all sets consisting each of one point are
convex.
The subsets of Rn defined below in 4. 5, and 6 are all convex sets
n
in R . This can be easily established by a direct verification of the
definition 3 of a convex set.
t It is obvious that the definition of a convex set would be unchanged if any of the other
line segments denned in 2 were used here instead of the closed line segment.

Fig. 3.1.1 Line and line segment through x1 and x*.

39
8.1 Nonlinear Programming

Fig. 3.1.2 Convex sets.

Halfspace
Let c G Rn, c 7* 0, and a G R- Then the set [x x G R", ex < a]
is an open half space in Rn, and the set {x \ x G #n, cz ^ a j is a c/oserf
halfspace in #". (Both halfspaces are convex sets.)
Plane
Letc G #n, c ?* 0, and a G #• Then the set {z | x G ^n, ex = a}
is called a plane in .Rn. (Each plane in Rn is a convex set.)

Subspace
A set F C Rn is a subspace if

Each subspace of /2n contains the origin and is a convex set. The
subspaces of R3 consist of 0, R3, the origin, and all straight lines and
planes passing through the origin.
Problem
(i) Show that each open or closed ball
Bt(x) = {x \ x G Rn, \\x - x\\ < 6} Bt(x) = {x | x G Rn, \\x - x\\ ^ e]

Fig. 3.1.3 Nonconvex sets.

iO
Convex Sets in R* 3.1

around a point x G Rn is a convex set. (Hint: Use the triangle inequality

1.3.10 in the form \\x + y\\ ^ \\x\\ + \\y\\.)
(ii) Show that the interior of a convex set is convex.

Vertex
Let F be a convex set in Rn. Each x G F for which there exist no
two distinct xl,xz G F different from x such that x G [z1,^2], is called a
vertex of F (or an extreme point of F).
A convex set F C Rn may have no vertices (for example the plane
{x | x G Rn, ex = a} and the open ball B\(f) have no vertices), a finite
number of vertices (for example the set {x \ x G Rn, x ^ 0, ex = 1},
where e in an n-vector of ones, has the n vertices e\ i = 1, . . . , n,
where & is an n-vector with ef = 1 and e? = 0, i 5^ j), or an infinite
number of vertices (for example the closed ball J?\(z)C Rn has an
infinite number of vertices given by [x \ x G R", \\x — x\\ = X}).

Theorem
// (Fi),-e/ is a family (finite or infinite] of convex sets in Rn, then their
intersection C\ Ft- is a convex set.
i&

PROOF Let x^x* G H Ft, and let 0 ^ X ^ 1. Then for each i G /,

iGI
x ,x G F,, and since F, is convex, (1 — \)xl + \x2 G Ft.
l 2

Polytope and polyhedron

A set in Rn which is the intersection of a finite number of closed
halfspaces in Rn is called a polytope. If a polytope is bounded (that is,
for each x in the polytope \\x\\ ^ a for some fixed a G R), it is called a
polyhedron.
It follows from the convexity of the halfspaces 4 and Theorem 9
that poly topes and polyhedra are convex sets.

Convex combination
A point b G Rn is said to be a convex combination of the vectors
0 , • • • , «TO G Rn if there exist m real numbers pit . . . , pm such that
1

b = pia1 + • • • + pmam, pi, . . . , pm ^ 0, pi + • • • 4- Pm = 1

Equivalently, if we define an m X n matrix A whose ith row is Ai = a',
41
3.1 Nonlinear Programming

and if we let p = (pit . . . ,pm) £ Rm and e be an m-vector of ones, then

we have that 6 is a convex combination of the rows of A if
(6 = A'p, p ^ 0, ep = 1) has a solution p E Rm
Note that if 6 is a convex combination of two points a1,a2 £ Rn,
then this is equivalent to saying that 6 £ [a^o2] (see 2).

Simplex
Let x°, xl, . . . , xm be m + 1 distinct points in Rn, with m ^ n.
If the vectors xl — x°, . . . , xm — x° are linearly independent, then the
set of all convex combinations of x°, xl, . . . , xm

is called an m-simplex in /2n with vertices x°, x1, . . . , xm. (A 0-simplex

is a point, a 1-simplex is a closed line segment, a 2-simplex is a triangle,
and a 3-simplex is a tetrahedron.)

Theorem
A set F C Rn i$ convex if and only if for each integer m ^ 1, every
convex combination of any m points of T is in T. Equivalently, a necessary
and sufficient condition for the set T to be convex is that for each integer m ^ 1

PROOF The sufficiency of 14 is trivial; take m — 2, then T is convex

by 3.
The necessity of 14 will be shown by induction. For m = 1, 14
holds trivially. For m — 2, 14 holds as a consequence of 3. Assume
now that 14 holds for m, we will now show that it also holds for m -f 1.
Let
x\ x*, . . . , xn+l £ T
Pit • • • , Pm+i ^ 0
Pi + • • • + Pm+1 = 1

If pm+i = 0, then pix1 + • • • -f pmxm £ F, since 14 holds for m. If

pm+1 = 1, thenpis 1 + • • • + Pm-nzm+1 = zm+1 £ r. If 0 < pm+1 < 1,
48
Convex Seta in Rn 3.1

then we can write

Caratheodory's theorem [Caratheodory 07]

Let F C Rn- If ^ is a convex combination of points of F, then x is a
convex combination of n + I or fewer points of F.

PROOF Let

We will show now that if m > n -f 1, then x can be written as a convex

combination of m — 1 points in F. (This would establish the theorem
then, for we could repeatedly apply the result until £ is a convex combina-
tion of n -f 1 points of F.) If any Pi in the above expression is zero,
then x is a convex combination of m — 1 or fewer points of T. So let
each pi > 0. Since m > n + 1, there exist rit . . . , rm_l G R, not all
zero, such that

Define
9» = Pi — otTi for i = I , . . . , m
where a is some positive number chosen such that #,- ^ 0 for all i, and
at least one g,-, say g*, is equal to 0. In particular we choose a such that

43
a.i Nonlinear Programming

Fig. 3.1.4 A set T and its convex hull [T].

Then

and

Hence x is a convex combination of m — 1 points in F.

Convex hull
Let T C Rn. The convex hull of T, denoted by [F], is the inter-
section of all convex sets in Rn containing F. (By Theorem 9, the convex
hull of any set F C Rn is convex. Figure 3.1.4 shows a hatched noncon-
vex set in Rz and its shaded convex hull.)
Obviously if F is convex, then F = [F].
Theorem
The convex hull [F] of a set F C Rn is equal to the set of all convex
combinations of points of F.

PROOF Let A denote the latter set, that is,

If x\x* G A, then

44
Conrex Sets in Rn 3.1

Hence for 0 g X ^ 1

and

Thus \xl -f- (1 — X)o;2 G A, and A is convex. It is also clear that r C A.

Since A is convex, then [T] C A. We also have by Theorem 13 that the
convex set [T] containing T must also contain all convex combinations of
points of I\ Hence A C [r], and A = [r].

Sum of two sets

Let T,A C Rn. Their sum r -f A is denned by

Product of a set with a real number

Let T C Rn, and let X E R- The product XT is defined by

Note that if X = -1 and r, A C Rn, then A + XF = A - T. Note that

this is not the complement of F relative to A as defined in 1.2 and written as
A ~ r.
Theorem
The sum Y + A of two convex sets T and A in Rn is a convex set.

PROOF Let 21,*2 E T + A, then zl = xl + yl and zz = x2 + yz, where

xl,x2 G T and y^y2 £ A. For 0 ^ X g 1

Hence r + A is convex.
Theorem
The product pY of a convex set T in Rn and the real number p is a
convex set.

45
1.2 Nonlinear Programming

PROOF Let zl,z* £ »T, then z1 = »xl, z2 = ^x\ where xl,x2 £ T. For
0 ^ X^ 1

Corollary
// F and A are two convex sets in R", then F — A is a convex set.

2. Separation theorems for convex sets

It is intuitively plausible that if we had two disjoint convex sets in Rn,
then we could construct a plane such that one set would lie on one
side of the plane and the other set on the other side. Despite its simplic-
ity, this is a rather deep result and is not easy to prove. One version of
this result, the Hahn-Banach theorem, can be established by only using
the vector space properties 1.3.3 of Rn and not the topological properties
induced by the norm ||xj! [Berge 03, Valentine 64]. We shall, however,
use these topological properties of Rn (all summarized in Appendix B) in
deriving the separation theorems for convex sets. In particular our
method of proof will make use of Gordan's theorem of the alternative
2.4-5 and the finite intersection theorem of compact sets B.3.2 (iii).
(Knowledge of the contents of Appendix B is assumed from here on.)

Separating plane
The plane {x \ x £ Rn, ex = a } , c 7^ 0, is said to separate (strictly
separate] two nonempty sets F and A in Rn if

If such a plane exists, the sets F and A are said to be separable (strictly
separable).
Figure 3.2.1 gives a simple illustration in R2 of two sets in Rn
which are separable, but which are neither disjoint nor convex. It should
be remarked that in general separability does not imply that the sets are
disjoint (Fig. 3.2.1], nor is it true in general that two disjoint sets are
separable (Fig. 3.2.2}. However, if the sets are nonempty, convex, and

46
Convex Sets In R? 3.2

Fig. 3.2.1 Separable but not disjoint sets.

disjoint, then they are separable, and in fact this is a separation theorem
we intend to prove.

Lemma
Let fl be a nonempty convex set in Rn, not containing the origin 0.
Then there exists a plane {x \x (~ Rn, ex = 0}, c 7* 0, separating £2 and 0,
thatis,

PROOF With every x G & we associate the nonempty closed set

Let x1, . . . , xm be any finite set of points in fl. It follows from the con-
vexity of fl, Theorem 3.1.13, and from the fact that 0 ^ 12, that

Fig. 3.2.2 Disjoint but not separable sets.

47
3.2 Nonlinear Programming

or equivalently

Hence by Gordan's theorem 2.4-5

x^y > 0, i — 1, . . . , m has a solution y G: Rn
Obviously y ^ 0, and we can take y such that yy = 1. Then

and hence

The sets (A^en are closed sets relative to the compact set {y \ y G Rn,
yy = 1} [see B.1.8 and B.3.2(\}}, hence by the finite intersection theorem
B.3.2(\\\) we have that C\ Ax ^ 0. Let c be any point in this intersec-
zen
tion. Then cc = 1 and cz ^ 0 for all x G ^- Hence {# | a: G #n,
ex = 0} is the required separating plane. |
It should be remarked that in the above lemma we did not impose
any conditions on fl other than convexity. The following example shows
that the above lemma cannot be strengthened to x G ^ =* ex > 0 without
some extra assumptions. The set

is convex and does not contain the origin, but there exists no plane
{x | x £ R*, ex = 0} such that x E & =» ex > 0 (Fig. 3.8.3).
If on the other hand we do assume that ft is closed (or even if we

Fig. 3.2.3

48
Convex Sets in Rn 3.9

assume less, namely that the origin is not a point of closure ft), then we
can establish a stronger result, that is, there exists a plane which strictly
separates the origin from ft (see Corollary 4 and Lemma 5 below). How-
ever, before doing this, we need to establish the following fundamental
separation theorem.
Separation theorem
Let F and A fee two nonempty disjoint convex sets in R". Then there
exists a plane {x | x G Rn, ex = a } , c ^ 0, which separates them, that is,

PROOF The set

is convex by Corollary 3.1.22, and it does not contain the origin 0 because
F C\ A = 0. By Lemma 2 above there exists a plane \x \ x G Rn, ex = 0},
c 7± 0, such that

or

Hence

Define

Then

We derive now from the above fundamental separation theorem a

corollary, and from the corollary a lemma, Lemma 5. Lemma 5 will be
used in establishing a strict separation theorem, Theorem 6, below.
Corollary
Let 12 be a nonempty convex set in Rn. If the origin 0 is not a point
of closure of fl (or equivalently if the origin is not in the closure fi of fi), then

49
3.2 Nonlinear Programming

there exists a plane {x \ x G Rn, ex = a ] , c 7* 0, a > 0, strictly separating

12 and 0, and conversely. In other words

PROOF («=) Assume that there exist c ^ 0, a > 0 such that ex > a
for all x G Q. If 0 G 12, then (see BJ.3 and B.I.6} there exists an x G f
such that ||z|| < a/2||c||, and hence

which is a contradiction. Hence 0 ££ 12.

(=>) Since 0 is not a point of closure of 12, there exists an open
ball Bf(0) = \x \ x G Rn, \\x\\ < e\ around 0 such that B e (0) n 0 = 0
(see B.1.3). Since the ball /?€(0) is convex (see 3.1.7), it follows by
Theorem 3 that there exists a plane {x \ x G Rn, ex = 7}, c 7* 0, such that

Since 5t(0) is an open ball, it must contain the nonzero vector Sc for
some positive 5. Hence 7 ^ dec > 0. Let a = ^dcc > 0. Then

Lemma
Let ft be a nonempty closed convex set in Rn. If 12 does not contain
the origin, then there exists a plane \x \ x G Rn, ex = a } , c ^ 0, a > 0,
strictly separating 12 and 0, and conversely. In other words

PROOF This lemma follows from Corollary 4 above by observing that

the requirement that 12 be closed and not contain the origin 0 implies
that 0 is not a point of closure of 12, that is, 0 ^ fl (see B.1.3, B.I.5 and
B.1.6).

Strict separation theorem

Let F and A be two nonempty convex sets in Rn, with T compact and
A closed. If T and A are disjoint, then there exists a plane {x \ x G Rn,

80
Convex Set* in R" 3.2

ex = a ] , c 5^ 0 which strictly separates them, and conversely In other

words

PROOF («=) If x E F C\ A, then ex < a < ex, a contradiction.

(=») The set

is convex by Corollary 3.1.22 and closed by Corollary B.3.8. Hence by

Lemma 5 above there exists a plane \x | x G Rn, ex = /x}, c 5^ 0, ^i > 0,
such that

orr

Hence

Define

Then

The above separation theorems will be used to derive some funda-

mental theorems for convex functions in the next chapter, which in turn
will be used in obtaining the fundamental Kuhn-Tucker saddlepoint
optimality criteria of convex nonlinear programming in Chap. 5 and also
the minimum principle necessary optimality condition of Chap. 11.
We remark here that a theorem of the alternative, the Gordan
theorem 8.4.6, was fundamental in deriving the above separation theo-
rems. We can reverse the process and use the above separation theorems
to derive theorems of the alternative. Thus to derive Gordan's theorem
2.4.5, namely that either A'y = 0, y > 0 has a solution y G Rm or Ax > 0

51
8.3 Nonlinear Programming

Fig. 3.2.4 Geometric reinterpretation of Gordan's theorem by using Lemma 6.

(a) A'y = 0, y > 0 has solution; Ax > 0 has no solution;
(b) Ax > 0 has solution; A'y = 0, y > 0 has no solution.

has a solution x G Rn, we observe that if e G Rm is a vector of ones, then

The last implication follows by taking y = &• G Rm, i = 1, . . . , m,

where & has zeros for all elements except 1 for the ith element.
Using the framework of Lemma 5 we can give a geometric reinter-
pretation of the Gordan's theorem as follows: Either the origin 0 £ R"

Fig. 3.2.5 Geometric interpretation of Lemma 5

02
Convex Bets in A" 3.2

is in the convex hull of the row vectors A I , . . . , An of the matrix A

(A'y = 0, y > 0 has a solution, Fig. 8.2.4a), or it is not (in which case
by Lemma 5 Ax > 0 has a solution x = c, Fig. 3.2.4V). More generally,
if fi is any nonempty closed convex set in Rn, then either it contains the
origin, Fig. 8.2.5a, or it does not (in which case by Lemma 5 there exists a
vector c £ Rn which makes a strict acute angle with each x G fy Fig.
3.2.5V).
Problem
Establish Farkas' theorem 2.4-6 by using Theorem 6 above.
(Hint: Observe that A'y = 6, y ^ 0 has no solution if and only if the sets
{6} and \z \ z = A'y, y ^ 0} are disjoint. Then use Theorem 6.)

us