Appendix A

Properties of positive (semi)definite matrices

In this appendix we list some well-known properties of positive (semi)definite matrices

which are used in this monograph. The proofs which are omitted here may be found
in [85]. A more detailed review of the matrix analysis which is relevant for SDP is
given by Jarre in [94].

A.1 CHARACTERIZATIONS OF POSITIVE (SEMI)DEFINITENESS

Theorem A.1 Let The following statements are equivalent:

(X is positive semidefinite);

All principal minors of X are nonnegative;

for some

We can replace positive semidefinite by positive definite in the statement of the

theorem by changing the respective nonnegativity requirements to positivity, and by
requiring that the matrix L in the last item be nonsingular. If X is positive definite
the matrix L can be chosen to be lower triangular, in which
case we call the Choleski factorization of X.
NB: In this monograph positive (semi)definite matrices are necessarily symmetric,
i.e. we will use positive (semi)definite instead of symmetric positive (semi)definite.1

1
In the literature a matrix is sometimes called positive (semi)definite if its symmetric part
is positive (semi)definite.
229
230 Aspects of Semidefinite Programming

As an immediate consequence of the second item in Theorem A.1 we have that

for any given, nonsingular

Another implication is that a block diagonal matrix is positive (semi)definite if and
only if each of its diagonal blocks is positive (semi)definite.

A.2 SPECTRAL PROPERTIES

The characterization of positive (semi)definiteness in terms of nonnegative eigenval-

ues follows from the RaleighRitz theorem.

Theorem A.2 (RaleighRitz) Let Then

It is well known that a symmetric matrix has an orthonormal set of eigenvectors,

which implies the following result.

Theorem A.3 (Spectral decomposition) Let Now A can be decomposed as

where is a diagonal matrix with the eigenvalues of A on the

diagonal, and Q is an orthogonal matrix with a corresponding set of orthonormal
eigenvectors of A as columns.

Since if we can define the symmetric square root

factorization of

Note that is the only matrix with this property.

Theorem A.4 Let and Then all the eigenvalues of XS are real
and positive.
 APPENDIX A: PROPERTIES OF POSITIVE (SEMI) DEFINITE MATRICES 231

Proof:
The proof is immediate by noting that

We will often use the notation

The eigenvalues of a symmetric matrix can be viewed as smooth functions on

in a sense made precise by the following theorem.

Theorem A.5 (Rellich) Let an interval be given. If is

a continuously differentiable function, then there exist continuously differentiable
functions such that give the values of
the eigenvalues of for each

The next lemma shows what happens to the spectrum of a positive semidefinite matrix
if a skew symmetric matrix is added to it, in the case where the eigenvalues of the sum
of the two matrices remain real numbers.

Lemma A.1 Let and let be skewsymmetric

One has Moreover, if then

Proof:
First note that is nonsingular since for all nonzero

using the skew symmetry of M. We therefore know that

since remains skewsymmetric. One now has that is a continuous function of

which is nowhere zero and strictly positive for as This shows

To prove the second part of the lemma, assume is such that

It then follows that By the same argument as above we then have
nonsingular, or

This implies that cannot be an eigenvalue of Similarly, cannot have

an eigenvalue smaller than This gives the required result.
232 Aspects of Semidefinite Programming

The spectral norm is the norm induced by the Euclidean vector

norm:

By the RaleighRitz theorem, the spectral norm and spectral radius coincide for
symmetric matrices:

The location of the eigenvalues of a matrix is bounded by the famous Gerschgorin

theorem. For symmetric matrices the theorem states that

As a consequence we find that the socalled diagonally dominant matrices are positive
semi-definite.

Theorem A.6 (Diagonally dominant matrix is PSD) A matrix is called di-

agonally dominant if

If A is diagonally dominant, then

A.3 THE TRACE OPERATOR AND THE FROBENIUS NORM

The trace of an matrix A is defined as

The trace is clearly a linear operator and has the following properties.

Theorem A.7 Let and Then the following holds:

The last item shows that we can view the usual Euclidean inner product on as an
inner product on
 APPENDIX A: PROPERTIES OF POSITIVE (SEMI)DEFINITE MATRICES 233

The inner product in (A.2) induces the so-called Frobenius (or Euclidean) norm on

It now follows from the first item in Theorem (A.7) that

The Frobenius and spectral norms are sub-multiplicative, i.e.

and
One can easily prove the useful inequality:

that is equivalent to

Inequality (A.4) follows from (A.3) by replacing A by and B by in (A.3).

Conversely, (A.3) follows from (A.4) by replacing A by and B by in (A.4).

Theorem A.8 (Fejer) A matrix is positive semidefinite if and only if

for all In other words, the cone is self-dual.

Proof:
Let and then

Conversely, if and for all then let be given and

set Now

For positive semidefinite matrices, the trace dominates the Frobenius norm, i.e.
234 Aspects of Semidefinite Programming

This follows by applying the inequality

to the nonnegative eigenvalues of X. Similarly, one can apply the arithmetic-geometric

mean inequality

to the eigenvalues of to obtain the inequality

where we have used the fact that for any

Lemma A.2 If and and then

Proof:
By the properties of the trace operator

Thus if it follows that Pre-multiplying by and post-

multiplying by yields which in turn implies

The following lemma is used to prove that the search directions for the interior
point methods described in this monograph are well defined. The proof given here is
based on a proof given by Faybusovich [54].

Lemma A.3 Let be linearly independent, and let

The matrix with entries

is positive definite.
Proof:
We prove that the quadratic form
 APPENDIX A: PROPERTIES OF POSITIVE (SEMI)DEFINITE MATRICES 235

is strictly positive for all nonzero To this end, note that for given

Denoting (which is nonzero by the linear independence of the

one has:

where the inequality follows from and

A.4 THE LWNER PARTIAL ORDER AND THE SCHUR

COMPLEMENT THEOREM

We define a partial ordering on via:

This partial ordering is called the Lwner partial order on (It motivates the alter-
native notation instead of It follows immediately that

One also has

The Schur complement theorem gives us useful ways to express positive semidefinite-
ness of matrices with a block structure.

Theorem A.9 (Schur complement) If

where A is positive definite and C is symmetric, then the matrix

is called the Schur complement of A in X. The following are equivalent:

M is positive (semi)definite;

is positive (semi)definite.
236 Aspects of Semidefinite Programming

Proof:
The result follows by setting and noting that

Since a block diagonal matrix is positive (semi)definite if and only if its diagonal
blocks are positive (semi)definite, the proof is complete.
Appendix B
Background material on convex optimization

In this Appendix we give some background material on convex analysis, convex opti-
mization and nonlinear programming. All proofs are omitted here but may be found
in the books by Rockafellar [160] (convex analysis) and Bazaraa et al. [16] (nonlinear
programming).

B.1 CONVEX ANALYSIS

Definition B.1 (Convex set) Let two points and be given.

Then the point

is a convex combination of the two points

The set is called convex, if all convex combinations of any two points
are again in

Definition B.2 (Convex function) A function defined on a convex set is

called convex if for all and one has

The function is called strictly convex if the last inequality is strict.

A function is convex if and only if its epigraph is convex.

Definition B.3 (Epigraph) The epigraph of a function is the

dimensional set

237
238 Aspects of Semidefinite Programming

Theorem B.1 A twice differentiable function is convex (resp. strictly convex) on

an open set if and only if its Hessian is positive semidefinite (resp. positive
definite) on

Example B.1 The function defined by has a

positive definite Hessian and is therefore strictly convex. (This is proven in Appendix
C.)

Strictly convex functions are useful for proving uniqueness properties, due to the
following result.

Theorem B.2 If a strictly convex function has a minimizer over a convex set, then this
minimizer is unique.

Definition B.4 (Convex cone) The set is a convex cone if it is a convex set
and for all and one has

Example B.2 Four examples of convex cones in are:

the symmetric positive semidefinite cone:

the copositive cone:

the cone of completely positive matrices:

the cone of nonnegative matrices:

Definition B.5 (Face (of a cone)) A subset of a convex cone is called a face of
if for all and one has if and only if and
 APPENDIX B: BACKGROUND MATERIAL ON CONVEX OPTIMIZATION 239

Example B.3 An example of a face of the cone of positive semidefinite matrices

is

Note that if and then if and only if and

Definition B.6 (Extreme ray (of a cone)) A subset of a convex cone is called an
extreme ray of if it is a one dimensional face of i.e. a face that is a half line
emanating from the origin.

Example B.4 Any defines an extreme ray of the cone of positive semidef-
inite matrices via

Similarly, any defines an extreme ray of the cone of completely positive

semidefinite matrices via

Definition B.7 Let a convex set be given. The point is in the relative interior
of if for all there exists and such that
The set of relative interior points of the set will be denoted by

Theorem B.3 Assume that and are nonempty convex sets and
Then

Example B.5 Note that the interior of the cone of positive semidefinite matrices
is the cone of positive definite matrices Let denote an affine space.
Assume that there exists an that is also positive definite. Then, by the last
theorem,

since
240 Aspects of Semidefinite Programming

Theorem B.4 (Separation theorem for convex sets) Let and be nonempty con-
vex sets in There exists a such that

and

if and only if the relative interiors of and are disjoint.

The second inequality merely excludes the uninteresting case where the separating
hyperplane contains both and i.e. it ensures so-called proper separation.

B.2 DUALITY IN CONVEX OPTIMIZATION

We consider the generic convex optimization problem:

s.t.

where is a convex set and are differentiable convex functions

on (or on an open set that contains the set ).
For the convex optimization problem (CO) one defines the Lagrange function (or
Lagrangian)

where and

Definition B.8 A vector pair and is called a saddle

point of the Lagrange function L if

for all and

One easily sees that (B.3) is equivalent with

Lemma B.1 The vector and is a saddle point of

if and only if
 APPENDIX B: BACKGROUND MATERIAL ON CONVEX OPTIMIZATION 241

Since we can reformulate (CO) as

it follows that is an optimal solution of problem (CO) if there exists a such

that is a saddle point of the Lagrangian.
To ensure the existence of a saddle point of the Lagrangian, it is sufficient to require
the so-called Slater regularity condition (Slater constraint qualification).

Assumption B.1 (Slater regularity) There exists an such that

if not linear or affine;

if is linear or affine.

Under the Slater regularity assumption, we therefore have a one-to-one correspon-

dence between a saddle point of the Lagrangian and an optimal solution of (CO).

Theorem B.5 (KarushKuhnTucker) The convex optimization problem (CO) is gi-

ven. Assume that the Slater regularity condition is satisfied. The vector is an optimal
solution of (CO) if and only if there is a vector such that is a saddle point of
the Lagrange function L.

The formulation of the saddle point condition in Lemma (B.1) motivates the con-
cept of the Lagrangian dual.

Definition B.9 (Lagrangian dual) Denote

The problem

is called the Lagrangian dual of the convex optimization problem (CO).

It is straightforward to show the so-called weak duality property.

Theorem B.6 (Weak duality) If is a feasible solution of (CO) and then

242 Aspects of Semidefinite Programming

and equality holds if and only if

Under the Slater regularity assumption we have a stronger duality result, by the Karush
KuhnTucker theorem and Lemma (B.1).

Theorem B.7 (Strong duality) Assume that (CO) satisfies the Slater regularity con-
dition, and let be a feasible solution of (CO). Now the vector is an optimal solution
of (CO) if and only if there exists a such that is an optimal solution of the
Lagrangian dual problem and

B.3 THE KKT OPTIMALITY CONDITIONS

We now state the KarushKuhnTucker (KKT) necessary and sufficient optimality

conditions for problem (CO). First we define the notion of a KKT point.

Definition B.10 (KKT point) The vector is called a KarushKuhn

Tucker (KKT) point of (CO) if

(i)

(ii)

(iii)

(iv)

A KKT point is a saddle point of the Lagrangian L of (CO). Conversely, a saddle point
of L, is a KKT point of (CO). This leads us to the following result.

Theorem B.8 (KKT conditions) If is a KKT point, then is an optimal so-

lution of (CO). Conversely under the Slater regularity assumption a feasible
solution of (CO) is optimal if there exists a such that is a KKT point.

We say that meets the KKT conditions if there exists a such that
is a KKT point of (CO).
If we drop the convexity requirements on and in the statement of (CO), then
the KKT conditions remain necessary optimality conditions under the Slater regularity
assumption.
Appendix C
The function log det(X)

In this appendix we develop the matrix calculus needed to derive the gradient and
Hessian of the function log det(X ), and show that it is a strictly concave function.

Lemma C.1 Let be given by

Denoting

one has
Proof:
Let be given and let be such that One has

By the arithmetic-geometric inequality applied to the eigenvalues of one

has

243
244 Aspects of Semidefinite Programming

Using the well-known inequality we arrive at

This shows that is a subgradient of at Since is assumed differentiable,

the subgradient is unique and equals the gradient

The proof of the next result is trivial.

Lemma C.2 Let be given by

where One has

The following result is used to derive the Hessian of the log-barrier function

Lemma C.3 Let be given by

If denotes the derivative of with respect to X, then is

the linear operator which satisfies

for a given invertible X.

Proof:
Let denote the space of linear operators which map to The Frechet
derivative of is defined as the (unique) function such that

We show that satisfies (C.1). To this end, let be

such that is invertible, and consider
 APPENDIX C: THE FUNCTION log det(X ) 245

which shows that (C.1) indeed holds.

By Lemma A.3, the Hessian of the function is a positive defi-

nite operator which implies that is strictly convex on We state this observation
as a theorem.

Theorem C.1 The function defined by

is strictly convex.

An alternative proof of this theorem is given in [85] (Theorem 7.6.6).

Appendix D
Real analytic functions

This appendix gives some elementary properties of analytic functions which are used
in this monograph. It is based on notes by M. Halick [76].

Definition D.1 A function is said to be analytic at if there exist

and such that

Remark D.1 Taking the nth derivative in (D.1), it is easy to see that
Hence the series in (D. 1) is the Taylor series of at

Remark D.2 A function that is analytic at is infinitely differentiable

in some neighborhood of a and the corresponding Taylor series converges to
in some neighborhood of a (sometimes this is used as the definition of an analytic
function). If a function is infinitely differentiable, then it is not necessarily analytic. A
well-known example is the Cauchy function for and
for At all derivatives are zero and hence the corresponding Taylor series
converges to the zero function.

Definition D.2 Let be an open interval. Then is analytic on I if it is analytic

at any point

We can extend this definition to closed intervals by making the following changes
to the above.

(a) if is defined for all then is defined to be analytic at a if there

exist an and a series of coefficients such that the equality in (D.1)
holds for all
247
248 Aspects of Semidefinite Programming

(b) if is defined for all then we say that can be analytically extended
to if there exist an and a series of coefficients such that the equality
in (D.1) holds for all

The following result is used in the proof of Theorem 3.6 (that the central path has
a unique limit point in the optimal set).

Theorem D.1 Let be analytic on [0,1). Let and be

an accumulation point of all such that Then for all

Proof:
Since is analytic at there exist and such that

We show that Let be the first non-zero coefficient in (D.2). Hence

Let The series in (D.3) converges at and hence the numbers

are bounded. Let i.e., Then

for all The last expression in (D.4) is positive for all sufficiently small
This is in contradiction with the assumption that has roots arbitrarily close to
Hence and in some right neighborhood of 0. Using the
analyticity in (0,1) it can be shown that it is zero on the whole domain.
Appendix E
The (symmetric) Kronecker product

This appendix contains various results about the Kronecker and symmetric Kronecker
product. The part on the symmetric Kronecker product is based on Appendix D in
Aarts[l].
We will use the vec and svec notation frequently in this appendix, and restate the
definitions here for convenience.

Definition E.1 For any symmetric matrix U, the vector

is defined as

such that

where

The inverse map of svec is denoted by smat.

E.1 THE KRONECKER PRODUCT

Definition E.2 Let and Then is defined as the

matrix with block structure

i.e. the block in position is given by

The following identities are proven in Horn and Johnson [86]. (We assume that the
sizes of the matrices are such that the relations are defined and that the inverses exist
where referenced.)
249
250 Aspects of Semidefinite Programming

Theorem E.1 Let K, L, G and H be real matrices.

The eigenvalues of are given by As a

consequence, if G and K are positive (semi)definite, then so is

If and then is an eigenvector of

with corresponding eigenvalue

As an application of the Kronecker product we can analyse the solutions of so-

called Lyapunov equations.

Theorem E.2 Let and The Lyapunov equation

has a unique symmetric solution if A and A have no eigenvalues in common.

Proof:
Using the first item in Theorem E.1, we can rewrite (E.1) as

By using the fourth item of Theorem E. 1 we have

In other words, the matrices and commute and therefore share a set of
eigenvectors. Moreover, by the fifth item of Theorem E. 1 we know that the eigenvalues
of and are obtained by taking copies of the spectrum of A. Therefore each
eigenvalue of is given by for some It follows
that the matrix is nonsingular if A and A have no eigenvalues in
common. In this case equation (E.1) has a unique solution. We must still verify that
this solution is symmetric. To this end, note that if X is a solution of (E.1), then so is
This completes the proof.
 APPENDIX E: THE (SYMMETRIC) KRONECKER PRODUCT 251

E.2 THE SYMMETRIC KRONECKER PRODUCT

Definition E.3 The symmetric Kronecker product of any two matrices G and
K (not necessarily symmetric), is a mapping on a vector where U is a
symmetric matrix and is defined as

Note that the linear operator is defined implicitly in (E.2). We can give a
matrix representation of by introducing the orthogonal matrix
(i.e. with the property that

Theorem E.3 Let be the unique orthogonal matrix that satisfies

(E.3). For any and one has

Proof:
Let be given and note that

where we have used the first identity in Theorem E. 1 to obtain the third equality.

Definition E.4 If for every vector where U is a symmetric nonzero

matrix,
252 Aspects of Semidefinite Programming

then is called positive definite.

Lemma E.1 The symmetric Kronecker product has the following properties.

1.
2.

3. for two symmetric matrices U and V;

4. If G and K are symmetric positive definite, then is positive definite;
5.

Proof:

1. This directly follows from Definition E.3.

2. Let U be a symmetric matrix, then

3. See Definition E.1.

4. For every symmetric nonzero matrix U we have to prove that

where Now

since G and K are symmetric. By using Property 3 we derive

Since U is nonzero and K and G are symmetric positive definite and therefore
nonsingular, we obtain that and thus
 APPENDIX E: THE (SYMMETRIC) KRONECKER PRODUCT 253

5. Define and for arbitrary symmetric matrices U and L

and Now

and by using Property 3 it follows that

Since we obtain

and from the definition of we derive

and thus

Definition E.5 For every nonsingular matrix P we define the operators and as

By using Property 2 in Lemma E. 1 we obtain

and

We define the operator for any nonsingular matrix P as

254 Aspects of Semidefinite Programming

The next theorems are based on Theorem 3.1 and Theorem 3.2 in Todd et al. [173].

Theorem E.4 If the matrices X and S are positive definite and the matrix
is symmetric positive semidefinite, then is positive definite.

Proof:
Consider an arbitrary nonzero vector We now prove that
Defining and where is also nonzero since exists, it
follows by using Property 5 in Lemma E. 1 that

By using Property 2 in Lemma E.1 we derive

Since X and S are symmetric positive definite, and are symmetric

positive definite and by using Property 4 in Lemma E. 1 it follows that
is positive definite. Therefore

By using Property 3 in Lemma E. 1 we now obtain

Since we assumed that is positive semidefinite and the matrix K is symmet-

ric, it follows that the matrix is positive semidefinite. Thus
 APPENDIX E: THE (SYMMETRIC) KRONECKER PRODUCT 255

Theorem E.5 If the matrices X and S are symmetric positive definite and the matri-
ces and commute, then is symmetric positive definite.

Proof:
If and commute, it follows that

Therefore the matrix is symmetric and

From Theorem A.4 we know that X S and therefore have positive

eigenvalues. Since is symmetric it is therefore positive definite.
Appendix F
Search directions for the embedding problem

Conditions for the existence and uniqueness of several search directions for the self
dual embedding problem of Chapter 4 (Section 4.2) are derived here.
The feasible directions of interior point methods for the embedding problem (4.4)
can be computed from the following generic linear system:

where and

where is the linear transformation given by

for any matrix M, and where the scaling matrix P determines the symmetrization
strategy. The best-known choices of P from the literature are listed in Table F. 1. We
will now prove (or derive sufficient conditions) for existence and uniqueness of the
search directions corresponding to each of the choices of P in Table F. 1. To this end,
257
258 Aspects of Semidefinite Programming

we will write the equations (F.1) and (F.2) as a single linear system and show that the
coefficient matrix of this system is nonsingular.1
We will use the notation

and define by replacing X by and S by in Table F.1. We will rewrite (F.2)

by using the symmetric Kronecker product. For a detailed review of the Kronecker
product see Appendix E; we only restate the relevant definitions here for convenience.

The symmetric Kronecker product of is implicitly defined

via

Using the symmetric Kronecker notation, we can combine (F.1) and (F.2) as

1
The approach used here is a straightforward extension of the analysis by Todd et al. in [173], where this
result was proved for SDP problems in the standard form (P) and (D).
 APPENDIX F: SEARCH DIRECTIONS FOR THE EMBEDDING PROBLEM 259

where

By Theorem E.4 in Appendix E we have the following result.

Lemma F.1 (Toh et al. [175]) Let be invertible and and symmetric positive
definite. Then the matrices E and F are invertible. If one also has
then the symmetric part of is also positive definite.

We are now in a position to prove a sufficient condition for uniqueness of the search
direction.

Theorem F.1 The linear system (F.3) has a unique solution if

Proof:
We consider the homogeneous system

and prove that it has only the zero vector as solution.

From (F.4) we have

and

Eliminating from the last two equations gives

System (F.4) also implies

260 Aspects of Semidefinite Programming

From (F.6) we have

The first term on the left-hand side is zero, by (F.7), and the second term is zero by the
skew-symmetry of We therefore have

which shows that since is assumed to be (non-symmetric) positive

definite. It follows that by (F.5). Furthermore, by (F.6), since has
full rank (the matrices are linearly independent).

All that remains is to analyze the condition

in the theorem. For the first three choices of in Table F.1, condition (F.8) always
holds (by Theorem E.5 in Appendix E). For (the so-called AHO direction),
(F.8) becomes the condition
An alternative sufficient condition for existence of the AHO direction was derived
by Monteiro and Zanjcomo [126], namely

where
Appendix G
Regularized duals

In this appendix we review some strong duality results for SDP problems that do not
satisfy the Slater condition.
The duals in question are obtained through a procedure called regularization. Al-
though a detailed treatment of regularization is beyond the scope of this monograph,
the underlying idea is quite simple:1 if the problem

is feasible, but not strictly feasible, we can obtain a strictly feasible reformulation by
replacing the semidefinite cone by a suitable lower dimensional face (say of
it, such that the new problem

is strictly feasible in the sense that there exists a pair such that and
Problem will have a perfect dual, by the conic duality
theorem (Theorem 2.3). The main point is to find an explicit expression of the dual of
the face The resulting dual problem now takes the form:

where denotes the dual cone of In the SDP case can be described by a
system of linear matrix inequalities. There is more than one way to do this and one
can obtain different dual problems within this framework (see Pataki [144] for details).

1
The idea of regularization was introduced by Borwein and Wolkowicz [30]. For a more recent (and sim-
plified) treatment the reader is referred to the excellent exposition by Pataki [144].

261
262 Aspects of Semidefinite Programming

Ramana [153] first obtained a regularized dual for (D);2 the so-called gap-free (or
extended LagrangeSlater) primal dual of (D) takes the form:

subject to

where the variables are and and

Note that the gap-free primal problem is easily cast in the standard primal form. More-
over, its size is polynomial in the size of (D). Unlike the Lagrangian dual (P) of (D),
has the following desirable features:

(Weak duality) If and is feasible for

then

(Dual boundedness) If (D) is feasible, its optimal value is finite if and only if
is feasible.
(Zero duality gap) The optimal value of equals the optimal value of (D)
if and only if both and (D) are feasible.
(Attainment) If the optimal value of (D) is finite, then it is attained by

The standard (Lagrangian) dual problem associated with is called the corrected
dual The pair and are now in perfect duality; see Ramana and
Freund[154].
Moreover, a feasible solution to (D) can be extracted from a feasible solution to
The only problem is that does not necessarily attain its supremum,
even if (D) does.

2
In fact, Ramana did not derive this dual via regularization initially; it was only shown subsequently that it
can be derived in this way in Ramana et al. [155].
 APPENDIX G: REGULARIZED DUALS 263

Example G.1 It is readily verified that the weakly infeasible problem (D) in Example
2.2 has a weakly infeasible corrected problem

The possible duality relations are listed in Table G.1. The optimal value of is
denoted by
References

[1] L.P. Aarts. Primal-dual search directions in semidefinite optimization. Masters

thesis, Delft University of Technology, Faculty of Information Technology and
Systems, Delft, The Netherlands, 1999.
[2] D.V. Alekseevskij, E.B. Vinberg, and A.S. Solodovnikov. Geometry of spaces
of constant curvature. In E.B. Vinberg, editor, Geometry II, volume 29 of Ency-
clopedia of Mathematical Sciences. Springer-Verlag, 1993.
[3] F. Alizadeh. Combinatorial optimization with interior point methods and semi
definite matrices. PhD thesis, University of Minnesota, Minneapolis, USA,
1991.
[4] F. Alizadeh. Interior point methods in semidefinite programming with applica-
tions to combinatorial optimization. SIAM Journal on Optimization, 5:1351,
1995.
[5] F. Alizadeh, J.-P.A. Haeberley, and M.L. Overton. Complementarity and non-
degeneracy in semidefinite programming. Mathematical Programming, Series
B, 77(2): 129162, 1997.
[6] F. Alizadeh, J.-P.A. Haeberley, and M.L. Overton. Primal-dual methods for
semidefinite programming: convergence rates, stability and numerical results.
SIAM Journal on Optimization, 8(3):746768, 1998.
[7] F. Alizadeh and S. Schmieta. Symmetric cones, potential reduction methods. In
H. Wolkowicz, R. Saigal, and L. Vandenberghe, editors, Handbook of semidefi-
nite programming, pages 195233. Kluwer Academic Publishers, Norwell, MA,
2000.
[8] E.D. Andersen. Finding all linearly dependent rows in large-scale linear pro-
gramming. Optimization Methods and Software, 6(3):219227, 1995.
[9] E.D. Andersen and K.D. Andersen. The MOSEK interior point optimizer for
linear programming: an implementation of the homogeneous algorithm. In
H. Frenk, K. Roos, T. Terlaky, and S. Zhang, editors, High performance op-
timization, pages 197232. Kluwer Academic Publishers, 2000.
265
266 Aspects of Semidefinite Programming

[10] E.D. Andersen, J. Gondzio, C. Mszros, and X. Xu. Implementation of

interior-point methods for large scale linear programs. In T. Terlaky, editor,
Interior point methods of mathematical programming, pages 189252. Kluwer,
Dordrecht, The Netherlands, 1996.
[11] M.F. Anjos and H. Wolkowicz. A strengthened SDP relaxation via a second
lifting for the Max-Cut problem. Special Issue of Discrete Applied Mathematics
devoted to Foundations of Heuristics in Combinatorial Optimization, to appear.
[12] K.M. Anstreicher and M. Fampa. A longstep path following algorithm for
semidefinite programming problems. In P. M. Pardalos and H. Wolkowicz, ed-
itors, Topics in Semidefinite and Interior-Point Methods, volume 18 of Fields
Institute Communications Series, pages 181196. American Mathematical So-
ciety, 1998.
[13] V. Balakrishnan and F. Wang. Sdp in systems and control theory. In H. Wolkow-
icz, R. Saigal, and L. Vandenberghe, editors, Handbook of semidefinite pro-
gramming, pages 421442. Kluwer Academic Publishers, Norwell, MA, 2000.
[14] G.P. Barker and D. Carlson. Cones of diagonally dominant matrices. Pacific
Journal of Mathematics, 57:1532, 1975.
[15] E.R. Barnes. A variation on Karmarkars algorithm for solving linear program-
ming problems. Mathematical Programming, 36:174182, 1986.
[16] M.S. Bazarraa, H.D. Sherali, and C.M. Shetty. Nonlinear Programming: Theory
and Algorithms. John Wiley and Sons, New York, 1993.
[17] R. Beigel and D. Eppstein. 3-Coloring in time a no-MIS algorithm.
In Proc. 36th IEEE Symp. Foundations of Comp. Sci., pages 444453, 1995.
[18] M. Bellare and P. Rogaway. The complexity of approximating a nonlinear pro-
gram. Mathematical Programming, 69:429441, 1995.
[19] R. Bellman and K. Fan. On systems of linear inequalities in hermitian matrix
variables. In V.L. Klee, editor, Convexity, Vol. 7 Proc. Symposia in Pure Math-
ematics, pages 111. Amer. Math. Soc., Providence, RI, 1963.
[20] A. Ben-Tal, L. El Ghaoui, and A.S. Nemirovski. Robustness. In H. Wolkowicz,
R. Saigal, and L. Vandenberghe, editors, Handbook of semidefinite program-
ming, pages 139162. Kluwer Academic Publishers, Norwell, MA, 2000.
[21] A. Ben-Tal and A.S. Nemirovski. Structural design. In H. Wolkowicz, R. Saigal,
and L. Vandenberghe, editors, Handbook of semidefinite programming, pages
443467. Kluwer Academic Publishers, Norwell, MA, 2000.
[22] S.J. Benson and Y. Ye. DSDP3: Dual scaling algorithm for general positive
semidefinite programming. Working paper, Computational Optimization Lab,
Dept. of Management Science, University of Iowa, Iowa City, USA, 2001.
[23] S.J. Benson, Y. Ye, and X. Zhang. Solving large-scale sparse semidefinite pro-
grams for combinatorial optimization. SIAM Journal on Optimization, 10:443
461, 2000.
 REFERENCES 267

[24] Lenore Blum, Mike Shub, and Steve Smale. On a theory of computation and
complexity over the real numbers: NP-completeness, recursive functions and
universal machines. Bull. Amer. Math. Soc. (N.S.), 21(1): 146, 1989.
[25] I.M. Bomze. On standard quadratic optimization problems. Journal of Global
Optimization, 13:369387, 1998.
[26] I.M. Bomze, M. Budinich, P.M. Pardalos, and M. Pelillo. The maximum clique
problem. In D.-Z. Du and P.M. Pardalos, editors, Handbook of Combinatorial
Optimization, volume suppl. Vol. A, pages 174. Kluwer, Dordrecht, 1999.
[27] I.M. Bomze and E. de Klerk. Solving standard quadratic optimization prob-
lems via linear, semidefinite and copositive programming. Technical Report TR
200103, Insitut fr Statistik und Decision Support Systems, Universitt Wien,
2001.
[28] I.M. Bomze, M. Dr, E. de Klerk, C. Roos, A. Quist, and T. Terlaky. On copos-
itive programming and standard quadratic optimization problems. Journal of
Global Optimization, 18:301320, 2000.
[29] R. Boppana and M.M. Halldrsson. Approximating maximum independent sets
by excluding subgraphs. Bit, 32:180196, 1992.
[30] J.M. Borwein and H. Wolkowicz. Regularizing the abstract convex program. J.
Math. Anal. Appl., 83(2):495530, 1981.
[31] S.E. Boyd, L. El Ghaoui, E. Feron, and V. Balakrishnan. Linear matrix inequal-
ities in system and control theory. SIAM Studies in Applied Mathematics, Vol.
15. SIAM, Philadelphia, USA, 1994.
[32] S. Burer and R.D.C. Monteiro. A Projected Gradient Algorithm for Solving
the Maxcut SDP Relaxation. Optimization Methods and Software, 15:175200,
2001.
[33] J.-D. Cho and M. Sarrafzadeh. Fast approximation algorithms on Maxcut,
and ordering for VLSI applications. IEEE Transactions on
Computers, 47(11):12531266, 1998.
[34] C. Choi and Y. Ye. Solving sparse semidefinite programs using the dual scaling
algorithm with an iterative solver. Working paper, Computational Optimization
Lab, Dept. of Management Science, University of Iowa, Iowa City, USA, 2000.
[35] S.A. Cook. The complexity of theorem proving procedures. In Proceedings of
the 3rd annual ACM symposium on the Theory of Computing, pages 151158,
1971.
[36] W. Cook, C.R. Coullard, and G. Turan. On the complexity of cutting plane
proofs. Discrete Applied Mathematics, 18:2538, 1987.
[37] B. Craven and B. Mond. Linear programming with matrix variables. Linear
Algebra Appl., 38:7380, 1981.
[38] and Semidefinite program-
ming methods for the traveling salesman problem. In Proceedings of the 7th
International IPCO conference, pages 126136, 1999.
268 Aspects of Semidefinite Programming

[39] E. de Klerk. Interior Point Methods for Semidefinite Programming. PhD thesis,
Delft University of Technology, Delft, The Netherlands, 1997.
[40] E. de Klerk and D.V. Pasechnik. On the performance guarantee of MAX-3-CUT
approximation algorithms. Manuscript, March, 2001.
[41] E. de Klerk and D.V. Pasechnik. Approximation of the stability number of a
graph via copositive programming. SIAM Journal on Optimization, to appear.
[42] E. de Klerk, D.V. Pasechnik, and J.P. Warners. On approximate graph colouring
and algorithms based on the Submitted to Journal of
Combinatorial Optimization, 2000.
[43] E. de Klerk, J. Peng, C. Roos, and T. Terlaky. A scaled Gauss-Newton primal
dual search direction for semidefinite optimization. SIAM Journal on Optimiza-
tion, 11:870888, 2001.
[44] E. de Klerk, C. Roos, and T. Terlaky. Initialization in semidefinite programming
via a selfdual, skewsymmetric embedding. OR Letters, 20:213221, 1997.
[45] E. de Klerk, C. Roos, and T. Terlaky. A short survey on semidefinite program-
ming. In W.K. Klein Haneveld, O.J. Vrieze, and L.C.M. Kallenberg, editors, Ten
years LNMB. Stichting Mathematisch Centrum, Amsterdam, The Netherlands,
1997.
[46] E. de Klerk, C. Roos, and T. Terlaky. Infeasiblestart semidefinite programming
algorithms via selfdual embeddings. In P. M. Pardalos and H. Wolkowicz, ed-
itors, Topics in Semidefinite and Interior-Point Methods, volume 18 of Fields
Institute Communications Series, pages 215236. American Mathematical So-
ciety, 1998.
[47] E. de Klerk, C. Roos, and T. Terlaky. On primaldual pathfollowing algorithms
in semidefinite programming. In F. Giannessi, S. Komlsi, and T. Rapcsk, ed-
itors, New trends in mathematical programming, pages 137157. Kluwer Aca-
demic Publishers, Dordrecht, The Netherlands, 1998.
[48] E. de Klerk, C. Roos, and T. Terlaky. Polynomial primaldual affine scaling al-
gorithms in semidefinite programming. Journal of Combinatorial Optimization,
2:5169, 1998.
[49] E. de Klerk and H. van Maaren. On semidefinite programming relaxations of
Submitted to Annals of Mathematics of Artificial Intelligence,
2000.
[50] E. de Klerk, H. van Maaren, and J.P. Warners. Relaxations of the satisfiabil-
ity problem using semidefinite programming. Journal of automated reasoning,
24:3765, 2000.
[51] D. den Hertog, E. de Klerk, and C. Roos. On convex quadratic approximation.
CentER Discussion Paper 2000-47, CentER for Economic Research, Tilburg
University, Tilburg, The Netherlands, 2000.
[52] J. Dieudonn. Foundations of Modern Analysis. Academic Press, New York,
1960.
 REFERENCES 269

[53] I.I. Dikin. Iterative solution of problems of linear and quadratic programming.
Doklady Akademii Nauk SSSR, 174:747748, 1967. (Translated in: Soviet Math-
ematics Doklady, 8:674675, 1967).
[54] L. Faybusovich. On a matrix generalization of affinescaling vector fields.
SIAM J. Matrix Anal. Appl., 16:886897, 1995.
[55] L. Faybusovich. Semidefinite programming: a pathfollowing algorithm for
a linearquadratic functional. SIAM Journal on Optimization, 6(4): 10071024,
1996.
[56] U. Feige and M. Goemans. Approximating the value of two prover proof sys-
tems with applications to MAX 2SAT and MAX DICUT. In Proc. Third Israel
Symposium on Theory of Computing and Systems, pages 182189, 1995.
[57] U. Feige and J. Kilian. Zero knowledge and the chromatic number. J. Comput.
System Sci., 57:187199, 1998.
[58] A.V. Fiacco and G.P. McCormick. Nonlinear programming: sequential un-
constrained minimization techniques. John Wiley & Sons, New York, 1968.
(Reprint: Volume 4 of SIAM Classics in Applied Mathematics, SIAM Publica-
tions, Philadelphia, USA, 1990).
[59] A. Frieze and M. Jerrum. Improved approximation algorithms for MAX
and MAX BISECTION. In Proc. 4th IPCO conference, pages 113, 1995.
[60] K. Fujisawa, M. Kojima, and K. Nakata. Exploiting sparsity in primaldual
interiorpoint methods for semidefinite programming. Mathematical Program-
ming, 79:235253, 1997.
[61] M.R. Garey and D.S. Johnson. Computers and intractability: a guide to the the-
ory of NPcompleteness. W.H. Freeman and Company, Publishers, San Fran-
sisco, USA, 1979.
[62] A. Genz. Numerical computation of multivariate normal probabilities. J. Comp.
Graph. Stat., 1:141149, 1992.
[63] PE. Gill, W. Murray, M.A. Saunders, J.A. Tomlin, and M.H. Wright. On
projected Newton barrier methods for linear programming and an equivalence
to Karmarkars projective method. Mathematical Programming, 36:183209,
1986.
[64] M. Goemans and F. Rendl. Combinatorial optimization. In H. Wolkowicz,
R. Saigal, and L. Vandenberghe, editors, Handbook of semidefinite program-
ming, pages 343360. Kluwer Academic Publishers, Norwell, MA, 2000.
[65] M.X. Goemans. Semidefinite programming in combinatorial optimization.
Math. Programming, 79(1-3, Ser. B):143161, 1997. Lectures on mathemat-
ical programming (ismp97).
[66] M.X. Goemans and D.P. Williamson. Improved approximation algorithms
for maximum cut and satisfiability problems using semidefinite programming.
Journal of the ACM, 42(6): 11151145, 1995.
270 Aspects of Semidefinite Programming

[67] M.X. Goemans and D.P. Williamson. Approximation algorithms for MAX 3-
CUT and other problems via complex semidefinite programming. In Proceed-
ings of the 33rd annual symposium on theory of computing (STOC 01), pages
443452. ACM, 2001.
[68] D. Goldfarb and K. Scheinberg. Interior point trajectories in semidefinite pro-
gramming. SIAM Journal on Optimization, 8(4):871886, 1998.
[69] A.J. Goldman and A.W. Tucker. Theory of linear programming. In H.W. Kuhn
and A.W. Tucker, editors, Linear inequalities and related systems, Annals of
Mathematical Studies, No. 38, pages 5397. Princeton University Press, Prince-
ton, New Jersey, 1956.
[70] C.C. Gonzaga. Path following methods for linear programming. SIAM Review,
34:167227, 1992.
[71] L.M. Graa Drummond and Y. Peterzil. The central path in smooth convex
semidefinite programs. Optimization, to appear.
[72] M. Grtschel, L. Lovsz, and A. Schrijver. Geometric Algorithms and Combi-
natorial Optimization. SpringerVerlag, Berlin, 1988.
[73] M. Grtschel, L.A. Lovsz, and A. Schrijver. Geometric algorithms and com-
binatorial optimization. Springer Verlag, Berlin, 1988.
[74] J. Gu, P.W. Purdom, J. Franco, and B.W. Wah. Algorithms for the satisfiability
(SAT) problem: a survey. In D. Du, J. Gu, and P.M. Pardalos, editors, Satisfiabil-
ity problem: Theory and applications, volume 35 of DIMACS series in Discrete
Mathematics and Computer Science. American Mathematical Society, 1997.
[75] A. Haken. The intractability of resolution. Theoretical Computer Science,
39:297308, 1985.
[76] M. Halick. Private communication, 2001.
[77] M. Halick. Analyticity of the central path at the boundary point in semidefinite
programming. European Journal of Operations Research, to appear.
[78] M. Halick, E. de Klerk, and C. Roos. On the convergence of the central path
in semidefinite optimization. SIAM Journal on Optimization, to appear.
[79] E. Halperin and U. Zwick. Approximation algorithms for MAX 4-SAT and
rounding procedures for semidefinite programs. In Proc. 7th IPCO, pages 202
217, 1999.
[80] J. Hstad. Some optimal inapproximability results. In Proc. 29th Ann. ACM
Symp. on Theory of Comp., pages 110, 1997.
[81] J. Hstad. Clique is hard to approximate within Acta Mathematica,
182:105142, 1999.
[82] B. He, E. de Klerk, C. Roos, and T. Terlaky. Method of approximate centers for
semidefinite programming. Optimization Methods and Software, 7:291309,
1997.
 REFERENCES 271

[83] C. Helmberg and F. Oustry. Bundle methods and eigenvalue functions. In

H. Wolkowicz, R. Saigal, and L. Vandenberghe, editors, Handbook of semidefi-
nite programming, pages 307337. Kluwer Academic Publishers, Norwell, MA,
2000.
[84] C. Helmberg, F. Rendl, R.J. Vanderbei, and H. Wolkowicz. An interior-point
method for semidefinite programming. SIAM Journal on Optimization, 6:342
361, 1996.
[85] R.A. Horn and C.R. Johnson. Matrix Analysis. Cambridge University Press,
1985.
[86] R.A. Horn and C.R. Johnson. Topics in Matrix Analysis. Cambridge University
Press, 1991.
[87] W.-Y. Hsiang. On infinitesimal symmetrization and volume formula for spheri-
cal or hyperbolic tetrahedrons. Quart. J. Math. Oxford, 39(2):463468, 1988.
[88] B. Jansen. Interior Point Techniques in Optimization. PhD thesis, Delft Univer-
sity of Technology, Delft, The Netherlands, 1995.
[89] B. Jansen, C. Roos, and T. Terlaky. The theory of linear programming : Skew
symmetric selfdual problems and the central path. Optimization, 29:225233,
1994.
[90] B. Jansen, C. Roos, and T. Terlaky. Interior point methods: a decade after Kar-
markar. A survey, with application to the smallest eigenvalue problem. Statistica
Neerlandica, 50, 1995.
[91] B. Jansen, C. Roos, and T. Terlaky. A polynomial primaldual Dikintype algo-
rithm for linear programming. Mathematics of Operations Research, 21:431
353, 1996.
[92] B. Jansen, C. Roos, and T. Terlaky. A family of polynomial affine scaling al-
gorithms for positive semidefinite linear complementarity problems. SIAM
Journal on Optimization, 7(1):126140, 1997.
[93] B. Jansen, C. Roos, T. Terlaky, and J.-Ph. Vial. Primaldual algorithms for linear
programming based on the logarithmic barrier method. Journal of Optimization
Theory and Applications, 83:126, 1994.
[94] F. Jarre. Convex analysis on symmetric matrices. In H. Wolkowicz, R. Saigal,
and L. Vandenberghe, editors, Handbook of semidefinite programming, pages
1327. Kluwer Academic Publishers, Norwell, MA, 2000.
[95] Jr. J.E. Dennis and H. Wolkowicz. Sizing and least change secant methods.
SIGNUM, 10:12911314, 1993.
[96] J. Jiang. A long step primaldual path following method for semidefinite pro-
gramming. OR Letters, 23(l,2):5362, 1998.
[97] V. Kann, S. Khanna, J. Lagergren, and A. Panconesi. On the hardness of ap-
proximating Max k-Cut and its dual. Chicago Journal of Theoretical Computer
Science, 1997(2), June 1997.
272 Aspects of Semidefinite Programming

[98] D. Karger, R. Motwani, and M. Sudan. Approximate graph coloring by semidef-

inite programming. In 35th Symposium on Foundations of Computer Science,
pages 213. IEEE Computer Society Press, 1994.
[99] H. Karloff and U. Zwick. A 7/8-approximation algorithm for MAX 3SAT? In
Proc. 38th FOCS, pages 406415, 1997.
[100] N.K. Karmarkar. A new polynomialtime algorithm for linear programming.
Combinatorica, 4:373395, 1984.
[101] L. Khachiyan. A polynomial time algorithm in linear programming. Soviet
Mathematics Doklady, 20:191194, 1979.
[102] S. Khanna, N. Linial, and S. Safra. On the hardness of approximating the
chromatic number. Combinatorica, 20:393415, 2000.
[103] D.E. Knuth. The sandwich theorem. The Electronic Journal of Combinatorics,
1:148, 1994.
[104] M. Kojima, N. Megiddo, T. Noma, and A. Yoshise. A unified approach to
interior point algorithms for linear complementarity problems, volume 538 of
Lecture Notes in Computer Science. Springer Verlag, Berlin, Germany, 1991.
[105] M. Kojima, M. Shida, and S. Shindoh. Local Convergence of Predictor-
Corrector Infeasible-Interior-Point Algorithms for SDPs and SDLCPs. Math-
ematical Programming, 80:129160, 1998.
[106] M. Kojima, M. Shida, and S. Shindoh. A note on the Nesterov-Todd and the
Kojima-Shindoh-Hara search directions in semidefinite programming. Opti-
mization Methods and Software, 11-12:4752, 1999.
[107] M. Kojima, M. Shidah, and S. Shindoh. A predictor-corrector interior
point algorithm for the semidefinite linear complementarity problem using the
Alizadeh-Haeberley-Overton search direction. SIAM Journal on Optimization,
9(2):444465, 1999.
[108] M. Kojima, S. Shindoh, and S. Hara. Interior point methods for the mono-
tone semidefinite linear complementarity problem in symmetric matrices. SIAM
Journal on Optimization, 7(1):88125, 1997.
[109] S. Kruk, M. Muramatsu, F. Rendl, R.J. Vanderbei, and H. Wolkowicz. The
Gauss-Newton direction in linear and semidefinite programming. Optimization
Methods and Software, 15(1): 127, 2001.
[110] J.B. Lasserre. An explicit exact SDP relaxation for nonlinear 0-1 programs. In
Proc. IPCO VIII, pages 293303, 2001.
[111] J.B. Lasserre. Global optimization with polynomials and the problem of mo-
ments. SIAM J. Optimization, 11:796817, 2001.
[112] M. Laurent. A comparison of the Sherali-Adams, Lovsz-Schrijver and
Lasserre relaxations for 0-1 programming. Technical report pna-r0108, CWI,
Amsterdam, The Netherlands, 2001.
[113] M. Laurent. Tighter linear and semidefinite relaxations for max-cut based on
the Lovsz-Schrijver lift-and-project procedure. SIAM Journal on Optimization,
to appear.
 REFERENCES 273

[114] A.S. Lewis and M.L. Overton. Eigenvalue optimization. Acta Numerica,
5:149190, 1996.
[115] L. Lovsz. On the Shannon capacity of a graph. IEEE Trans. on Information
Theory, 25:17, 1979.
[116] L. Lovsz and A. Schrijver. Cones of matrices and setfunctions and 01 opti-
mization. SIAM Journal on Optimization, 1(2):166190, 1991.
[117] Z.-Q. Luo, J.F. Sturm, and S. Zhang. Superlinear convergence of a Symmet-
ric primaldual path following algorithm for semidefinite programming. SIAM
Journal on Optimization, 8(1):5981, 1998.
[118] Z.-Q. Luo, J.F. Sturm, and S. Zhang. Conic convex programming and self-dual
embedding. Optimization Methods and Software, 14(3): 196218, 2000.
[119] I.J. Lustig, R.E. Marsten, and D.F. Shanno. Interior point meth-
ods : Computational state of the art. ORSA Journal on Computing, 6:115, 1994.
[120] S. Mahajan and H. Ramesh. Derandomizing semidefinite programming based
approximation algorithms. In Proc. of the 36th Annual IEEE Symposium on
Foundations of Computer Science, pages 162169. IEEE, 1995.
[121] S. Mehrotra. On the implementation of a (primaldual) interior point method.
SIAM Journal on Optimization, 2:575601, 1992.
[122] J. Milnor. Singular points of complex hypersurfaces. Annals of mathematics
studies. Princeton University Press, Princeton, New Jersey, 1968.
[123] S. Mizuno, M.J. Todd, and Y. Ye. On adaptive step primaldual interior
point algorithms for linear programming. Mathematics of Operations Research,
18:964981, 1993.
[124] R.D.C. Monteiro. Primal-dual path-following algorithms for semidefinite pro-
gramming. SIAM Journal on Optimization, 7(3):663678, 1997.
[125] R.D.C. Monteiro, I. Adler, and M.G.C. Resende. A polynomialtime primal
dual affine scaling algorithm for linear and convex quadratic programming and
its power series extension. Mathematics of Operations Research, 15:191214,
1990.
[126] R.D.C. Monteiro and P.R. Zanjcomo. A note on the existence of the Alizadeh-
Haeberley-Overton direction in semidefinite programming. Mathematical Pro-
gramming, 78(3):393397, 1997.
[127] R.D.C. Monteiro and P.R. Zanjcomo. Implementation of primal-dual methods
for semidefinite programming based on Monteiro and Tsuchiya directions and
their variants. Optimization Methods and Software, 11/12:91140, 1999.
[128] R. Motwani and P. Raghavan. Randomized Algorithms. Cambridge University
Press, 1995.
[129] T.S. Motzkin and E.G. Straus. Maxima for graphs and a new proof of a theorem
of Turn. Canadian J. Math., 17:533540, 1965.
[130] M. Muramatsu. Affine scaling algorithm fails for semidefinite programming.
Mathematical Programming, 83(3):393406, 1998.
274 Aspects of Semidefinite Programming

[131] M. Muramatsu and R.J. Vanderbei. Primal-dual affine-scaling algorithm fails

for semidefinite programming. Mathematics of Operations Research, 24:149
175, 1999.
[132] K.G. Murty and S.N. Kabadi. Some NP-complete problems in quadratic and
linear programming. Mathematical Programming, 39:117129, 1987.
[133] A. Nemirovskii and P. Gahinet. The projective method for solving linear matrix
inequalities. Mathematical Programming, Series B, 77(2): 163190, 1997.
[134] Y.E. Nesterov. Global quadratic optimization on the sets with simplex structure.
Discussion paper 9915, CORE, Catholic University of Louvain, Belgium, 1999.
[135] Yu. Nesterov. Longstep strategies in interior point potential reduction meth-
ods. Mathematical Programming, Series B, 76(l):4794, 1997.
[136] Yu. Nesterov. Quality of semidefinite relaxation for nonconvex quadratic op-
timization. Discussion paper 9719, CORE, Catholic University of Louvain,
Belgium, March 1997.
[137] Yu. Nesterov and A.S. Nemirovski. Interior point polynomial algorithms in
convex programming. SIAM Studies in Applied Mathematics, Vol. 13. SIAM,
Philadelphia, USA, 1994.
[138] Yu. Nesterov and M.J. Todd. Selfscaled barriers and interiorpoint methods
for convex programming. Mathematics of Operations Research, 22(1): 142,
1997.
[139] Yu. Nesterov, H. Wolkowicz, and Y. Ye. Semidefinite programming relaxations
of nonconvex quadratic optimization. In H. Wolkowicz, R. Saigal, and L. Van-
denberghe, editors, Handbook of semidefinite programming, pages 361419.
Kluwer Academic Publishers, Norwell, MA, 2000.
[140] J.A. Olkin. Using semidefinite programming for controller design in active
noise control. SIAG/OPT Views and News, 8:15, Fall 1996.
[141] P.A. Parillo. Structured Semidefinite Programs and Semi-algebraic Ge-
ometry Methods in Robustness and Optimization. PhD thesis, California
Institute of Technology, Pasadena, California, USA, 2000. Available at:
http://www.cds.caltech.edu/~pablo/.
[142] D.V. Pasechnik. Bipartite sandwiches: bounding the size of a
maximum biclique. ArXiv e-print, 1999. Available online at
http://xxx.lanl.gov/abs/math.CO/9907109.
[143] G. Pataki. On the rank of extreme matrices in semidefinite programs and
the multiplicity of optimal eigenvalues. Mathematics of Operations Research,
23(2):339358, 1998.
[144] G. Pataki. A Simple Derivation of a Facial Reduction Algorithm, and Extended
Dual Systems. Technical Report, Columbia University, New York, USA, 2000.
[145] G. Pataki. The geometry of semidefinite programming. In H. Wolkowicz,
R. Saigal, and L. Vandenberghe, editors, Handbook of semidefinite program-
ming, pages 2966. Kluwer Academic Publishers, Norwell, MA, 2000.
 REFERENCES 275

[146] G. Plya. ber positive Darstellung von Polynomen. Vierteljschr. Naturforsch.

Ges. Zrich, 73:141145, 1928. (also Collected Papers, Vol. 2, 309313, MIT
Press, Cambridge, Mass., London 1974).
[147] L. Porkolab and L. Khachiyan. On the complexity of semidefinite programs.
Journal of Global Optimization, 10:351365, 1997.
[148] F.A. Potra and R. Sheng. A superlinearly convergent primaldual infeasible
interiorpoint algorithm for semidefinite programming. SIAM Journal on Opti-
mization, 8(4): 10071028, 1998.
[149] F.A. Potra and R. Sheng. On homogeneous interiorpoint algorithms for
semidefinite programming. Optimization methods and software, 9:161184,
1998.
[150] V. Powers and B. Reznick. A new bound for Plyas Theorem with applications
to polynomials positive on polyhedra. In Proceedings of MEGA 2000, 2000. To
appear in J. Pure and Applied Algebra.
[151] W.R. Pulleyblank. The combinatorial optimization top 10 list. Plenary talk at
the ISMP2000 conference, Atlanta, August 7-11, 2000.
[152] A.J. Quist, E. de Klerk, C. Roos, and T. Terlaky. Copositive relaxation for
general quadratic programming. Optimization Methods and Software, 9:185
209, 1998.
[153] M. Ramana. An exact duality theory for semidefinite programming and its
complexity implications. Mathematical Programming, Series B, 77(2): 129
162, 1997.
[154] M. Ramana and R. Freund. On the ELSD duality theory for SDP. Technical re-
port, Center of Applied Optimization, University of Florida, Gainsville, Florida,
USA, 1996.
[155] M. Ramana, L. Tunel, and H. Wolkowicz. Strong duality for semidefinite
programming. SIAM Journal on Optimization, 7(3):641662, 1997.
[156] M.V. Ramana and P.M. Pardalos. Semidefinite programming. In T. Terlaky,
editor, Interior point methods of mathematical programming, pages 369398.
Kluwer, Dordrecht, The Netherlands, 1996.
[157] J. Renegar. Condition numbers, the barrier method, and the conjugate-gradient
method. SIAM Journal on Optimization, 6(4): 879912, 1996.
[158] J. Renegar. A mathematical view of interior-point methods in convex optimiza-
tion. SIAM, Philadelphia, PA, 2001.
[159] Bruce Reznick. Some Concrete Aspects of Hilberts 17th Problem. Preprint 98-
002, Department of Mathematics, University of Illinois at Urbana-Champaign,
Urbana, Illinois, USA, 1998.
[160] R.T. Rockafellar. Convex analysis. Princeton University Press, Princeton, New
Jersey, 1970.
[161] C. Roos, T. Terlaky, and J.-Ph. Vial. Theory and Algorithms for Linear Opti-
mization: An interior point approach. John Wiley & Sons, New York, 1997.
276 Aspects of Semidefinite Programming

[162] C. Roos and J.-Ph. Vial. A polynomial method of approximate centers for linear
programming. Mathematical Programming, 54:295305, 1992.
[163] Alexander Schrijver. A comparison of the Delsarte and Lovsz bounds. IEEE
Trans. Inform. Theory, 25(4):425429, 1979.
[164] H.D. Sherali and W.P. Adams. A hierarchy of relaxations between the contin-
uous and convex hull representations for 01 programming problems. SIAM
Journal on Discrete Mathematics, 3(3):411430, 1990.
[165] M. Shida, S. Shindoh, and M. Kojima. Existence of search directions in
interiorpoint algorithms for the SDP and the monotone SDLCP. SIAM Journal
on Optimization, 8(2):387396, 1998.
[166] N.Z. Shor. Quadratic optimization problems. Soviet Journal of Computer and
System Sciences, 25:111, 1987.
[167] J.F. Sturm. Using SeDuMi 1.02, a MATLAB toolbox for optimization over
symmetric cones. Optimization Methods and Software, 11-12:625653, 1999.
[168] J.F. Sturm. Error bounds for linear matrix inequalities. SIAM Journal on Opti-
mization, 10(4): 12281248, 2000.
[169] J.F. Sturm and S. Zhang. On the long step pathfollowing method for semidef-
inite programming. OR Letters, 22:145150, 1998.
[170] J.F. Sturm and S. Zhang. Symmetric primaldual path following algorithms
for semidefinite programming. Applied Numerical Mathematics, 29:301316,
1999.
[171] K. Tanabe. Centered Newton method for mathematical programming. In Sys-
tem Modelling and Optimization: Proceedings of the 13th IFIPConference,
Berlin, August/September 1987, volume 113 of Lecture Notes in Control and
Information Sciences, pages 197206. Springer-Verlag, New York, 1988.
[172] M.J. Todd. A study of search directions in primal-dual interior-point methods
for semidefinite programming. Optimization Methods and Software, 11:146,
1999.
[173] M.J. Todd, K.C. Toh, and R.H. Ttnc. On the Nesterov-Todd direction
in semidefinite programming. SIAM Journal on Optimization, 8(3):769796,
1998.
[174] M.J. Todd and Y. Ye. A centered projective algorithm for linear programming.
Mathematics of Operations Research, 15:508529, 1990.
[175] K.C. Toh, M.J. Todd, and R.H. Ttnc. SDPT3 a Matlab software package
for semidefinite programming. Optimization Methods and Software, 11:545
581, 1999.
[176] L. Tunel. Potential reduction and primaldual methods. In H. Wolkowicz,
R. Saigal, and L. Vandenberghe, editors, Handbook of semidefinite program-
ming, pages 235266. Kluwer Academic Publishers, Norwell, MA, 2000.
[177] A. Urquhart. Open problem posed at SAT98 workshop, May 1014, Pader-
born, Germany, 1998.
 REFERENCES 277

[178] H. van Maaren. Elliptic approximations of propositional formulae. Discrete

Applied Mathematics, 96-97:223244, 1999.
[179] L. Vandenberghe and S. Boyd. SP: Software for semidefinite programming.
Users Guide, Stanford University, Stanford, USA, 1994.
[180] L. Vandenberghe and S. Boyd. A primaldual potential reduction algorithm
for problems involving matrix inequalities. Math. Programming, 69:205236,
1995.
[181] L. Vandenberghe and S. Boyd. Semidefinite programming. SIAM Review,
38:4995, 1996.
[182] R.J. Vanderbei, M.S. Meketon, and B.A. Freedman. A modification of Kar-
markars linear programming algorithm. Algorithmica, 1:395407, 1986.
[183] H. Wolkowicz, R. Saigal, and L. Vandenberghe, editors. Handbook of semidef-
inite programming. Kluwer Academic Publishers, Norwell, MA, 2000.
[184] S.J.Wright. Primaldual interior point methods. SIAM, Philadelphia, 1997.
[185] X. Xu, P.-F. Hung, and Y. Ye. A simplified homogeneous and selfdual linear
programming algorithm and its implementation. Annals of OR, 62:151171,
1996.
[186] H. Yang, Y. Ye, and J. Zhang. An approximation algorithm for the two-parallel
machines scheduling problem with capacity constraints. Manuscript, Depart-
ment of Management Sciences, Henry B. Tippie College of Business Adminis-
tration, University of Iowa, 2000.
[187] Y. Ye. Toward probabilistic analysis of interiorpoint algorithms for linear
programming. SIAM Journal on Optimization, 19:3852, 1994.
[188] Y. Ye. Interior point algorithms. Discrete Mathematics and Optimization.
Wiley-Interscience, New York, 1997.
[189] Y. Ye. Approximating quadratic programming with bound and quadratic con-
straints. Mathematical Programming, 84:219226, 1999.
[190] Y. Ye. A .699 approximation algorithm for max-bisection. Mathematical Pro-
gramming, 90:101111, 2001.
[191] Y. Ye, M.J. Todd, and S. Mizuno. An homogeneous and
selfdual linear programming algorithm. Mathematics of Operations Research,
19:5367, 1994.
[192] Y. Zhang. On extending some primal-dual interior point algorithms from linear
programming to semidefinite programming. SIAM Journal on Optimization,
8(2):365386, 1998.
[193] Q. Zhao, S.E. Karisch, F. Rendl, and H. Wolkowicz. Semidefinite programming
relaxations for the quadratic assignment problem. Journal of Combinatorial
Optimization, 2(1):71109, 1998.
[194] U. Zwick. Outward rotations: a new tool for rounding solutions of semidefinite
programming relaxations, with applications to MAX CUT and other problems.
In Proc. 31st STOC, pages 679687, 1999.
 Index

(D), 2, 22 analytic function, 247

(P), 2, 22 approximate graph colouring, 183
262 approximation algorithm, 174
B, 50 for MAX-3-SAT, 223
N, 50 for 175
T, 50 approximation guarantee, 174
134 for global optimization
136, 137 implementable, 208
35 in the weak sense, 207
22 for MAX-3-SAT, 228
22 for 182
35 for standard quadratic optimization, 208
22 for the max. stable set problem, 188
22 arithmeticgeometric mean inequality, 52, 53,
35 136, 234
117 arithmetic-geometric inequality, 243
89 arrow matrix, 4
78
12
barrier parameter, 76
115 big-M method, 16, 61
85, 123 bit model of computation, 17
adaptive, 87
block diagonal matrix, 230
43, 45, 140
Boolean quadratic (in)equality, 214
smat, 249
svec, 249
169 Cauchy function, 247
CauchySchwartz inequality, 88, 108, 137
active constraint, 36 centering component, 81
adjacency matrix, 161 centering parameter, 76
of the pentagon, 161 central path, 13, 42, 52, 102
affinescaling, 15 analyticity of, 46
primal, 81, 82 convergence rate of, 58
AHO direction, 260 derivatives of, 48
algebraic set, 56 limit point, 48
analytic center, 51 of the embedding problem, 67
of the duality gap level sets, 52 quadratic convergence to, 84, 120, 122
analytic curve tangential direction to, 48
central path as an, 46 centrality conditions, 41
analytic extension, 248 centrality function, 78
of the central path, 48 102
279
 280 Aspects of Semidefinite Programming

of Faybusovich, 78 epigraph, 151, 237

of Jiang, 117 extended LagrangeSlater dual, 262
Choleski factorization, 224, 229 extreme point, 190
chromatic number, 4, 158, 223 extreme ray, 190, 239
clause, 7, 212
clausevariable matrix, 214 face, 261
clique, 4, 222 of a convex cone, 238
clique number, 4, 158, 223 of the cone 34
coclique, 7 feasible direction, 24, 98
coclique number, 7 feasible set, 22
combinatorial optimization, 3 feasible step, 24
complementary graph, 158 Frobenius norm, 2, 44, 233
complementary solutions, 33 full NT step, 118
condition number
of the embedding, 70 gamma function, 183
cone gap relaxation, 221
of completely positive matrices, 189 gapfree primal problem, 262
extreme ray of, 190, 239 generalised eigenvalues, 135
of copositive matrices, 189 GoldmanTucker theorem, 33
of nonnegative matrices, 189 gradient
of positive semidefinite matrices, 12 of log det (X), 243
extreme ray of, 239 Gram matrix, 178
face of, 239
conic duality theorem, 28 heptagon, 167
conic problem formulation, 11, 24 Hessian
conjugate gradient method, 93 of log det(X), 244
conjunctive normal form (CNF), 212 homogeneous embedding, 16, 64
convex cone, 238
convex function, 149, 237 icosahedron, 201
convex set, 237 idempotent matrix, 177
copositive matrix, 187 illposedness, 71
curve selection lemma, 56 implicit function theorem, 46, 48
cutting plane, 220 improving ray, 29, 69
cycle (in a graph), 203 incidence vector, 158, 221
inclusionexclusion, 225
independence number, 8
damped NT step, 118, 124
independent set, 7
derandomization, 224
infeasibility, 22
defect edge, 169, 222
infeasiblestart method, 16
degree of a vertex, 185
inner iteration, 92, 124
deterministic rounding, 224
integer programming, 214
diagonally dominant, 172, 232
interior point assumption, 23
dihedral angle, 226
interior point methods, 11
Dikin ellipsoid, 82, 100
Dikintype primaldual affinescaling, 99 Jacobian, 47
directional derivative
of 137 KZ relaxation, 217, 228
of 137 KKT conditions, 242
dual barrier function, 76 KKT point, 228, 242
dual cone, 12, 178 Kronecker product, 47
dual degeneracy, 37 properties of, 249
duality gap, 2, 25, 99, 137
after full NT step, 119 Lwner partial order, 2, 235
compact level set of, 43 Lagrangian, 240
upper bound in terms of 137 Lagrangian dual, 2, 241
of (P), 22
ellipsoid method, 11, 17 of SDP in conic form, 24
elliptic representations of clauses, 215 Laplacian, 6, 93
 INDEX 281

legal colouring, 170, 183 optimization software, 18

liftandproject method, 8 CPLEX, 63
linear programming (LP), 3, 11 DSDP, 93
logarithmic barrier methods, 12 MOSEK, 63
logarithmic Chebychev approximation, 9 SDTP3, 15, 131
logical OR operator, 211 SeDuMi, 15, 64, 131, 154
logical variables, 211 SP, 146
long step method, 15 XPRESSMP, 63
dual algorithm, 92 orthogonal projection, 79
primaldual algorithm with NT direction, orthogonality property, 24, 67
124 outer iteration, 92, 124
Lorentz cone, 154
Lovsz 4, 158, 172, 189 Plyas theorem, 196
Lovsz sandwich theorem, 158, 223 pathfollowing method, 13
Lyapunov equation, 116, 131, 250 primal, 75
primaldual, 115
machine scheduling, 8 pentagon
matrix of, 161
completely positive, 189 approximations of the stability number of,
copositive, 10, 189 200
diagonally dominant, 232 Shannon capacity of, 167
idempotent, 177 perfect duality, 25, 73
nonnegative, 189 perfect graph, 162
positive semidefinite, 2 Petersen graph, 169
skewsymmetric, 105 maximum stable set of, 188
MAX-2-SAT, 7 pigeonhole formula, 218, 222
MAX-3-SAT, 7 plane search, 135, 136
approximation guarantee for, 228 pointed cone, 11
169 positive semidefiniteness
212 characterizations of, 229
MAX-BISECTION, 7 potential reduction method, 16, 133
MAX-CUT, 5 by Nesterov and Todd, 142
general framework, 134
maximal complementarity
of optimal solutions, 34 predictorcorrector method, 15, 115, 146
MizunoToddYe type, 126
of the limit point of the central path, 48,
49 primal barrier function, 42, 76
primal degeneracy, 36
maximum satisfiability problem, 7
primaldual affinescaling, 100, 109
maximum stable set problem, 188
primaldual barrier function, 13, 76, 117, 124
reduction from SAT, 212
primaldual Dikin ellipsoid, 99
multinomial coefficient, 197
primaldual Dikin-type direction, 136
multinomial theorem, 197
primaldual methods, 13
mutilated chess board formula, 219, 222
projected Newton direction, 78, 89
prepositional formula, 212
negation (of a logical variable), 211
Nesterov-Todd (NT) direction, 116 quadratic approximation, 149
Nesterov-Todd (NT) scaling, 98 multivariate case, 152
nondefect edge, 5 univariate case, 150
NPcomplete, 212 quadratic assignment problem, 8
quadratic least squares, 149
optimal partition, 35 quadratic programming (QP), 3
optimal solutions, 22
uniqueness of, 37, 38 Raleigh-Ritz theorem, 230
optimal value randomized algorithm, 171, 174
of SDP problem, 22 for 175
optimality conditions, 13, 42, 242 for satisfiable instances of MAX-3-SAT,
for (P) and ( D ) , 33 224
for projected Newton direction, 79 randomized rounding, 224
282 Aspects of Semidefinite Programming

for 2-clauses, 227 spherical tetrahedron, 227, 228

for 3clauses, 227 spherical triangle, 226
regularization, 29, 73, 261 angular excess of, 226
relative interior, 239 area of, 226
of the optimal set, 34 stability number, 187
relative volume, 228 stable set, 7, 188, 213
resolution, 220 stable set polytope, 8
robustness, 11 standard quadratic optimization, 204
copositive programming formulation of,
sandwich theorem, 172 205
satisfiability problem (SAT), 7, 211 LP approximation of, 205
scaled primaldual direction, 98 naive approximation of, 205
scaling matrix, 257 SDP approximation of, 207
Schrijvers 201 step length, 24
Schur complement, 235 strict complementarity, 33, 34, 38, 128
Schur complement theorem, 216, 235 strict feasibility, 23
SDP problems strictly convex function, 237
complexity of, 17 characterization of, 238
standard form, 22 minimizers of, 238
SDTP3, 15, 131 strong duality, 3, 25, 242
search directions, 67 strong feasibility, 23
second order cone, 3 strong infeasibility
second order solution characterization of, 29
of centering system, 130 strong product for graphs, 164
SeDuMi, 15, 64, 131, 154 structural design, 11
selfdual, 65, 66 sum of squares, 10
selfdual embedding, 16 superlinear convergence, 15, 128
extended formulation, 64 symmetric Kronecker product, 251
search directions for, 257 properties of, 252
self-concordant barrier, 12 symmetric square root factorization, 230
semicolouring, 184 symmetrization strategy, 14
semidefinite feasibility problem, 18 system and control theory, 10, 133
complexity of, 18
semidefinite programming, 1 TanabeToddYe potential function 16,
applications in approximation theory, 8 134, 137
applications in combinatorial optimiza- Taylor expansion, 247
tion, 4 trace, 2, 44, 99, 232
applications in engineering, 10 properties of, 232
interior point algorithms for, 11 traveling salesman problem, 8
review papers on, 18 triangle inequalities, 215
special cases of, 3 trianglefree graph, 203
separation theorem for convex sets, 26, 151, 239 truss, 11
sequential minimization methods, 12 truth assignment, 211, 224
Shannon capacity, 165 Turing machine, 17
Shor relaxation, 215
simple graph, 169 unboundedness, 22
skewsymmetric matrix, 105 unit hypersphere, 178, 227
Slater regularity, 23, 241
solvability, 22 weak duality, 2, 241
SP, 146 weak infeasibility, 29
sparsity, 101 characterization of, 31
spectral bundle method, 17 weakly improving ray, 31, 32
spectral decomposition, 48 weight of a cut, 5, 169
of optimal solutions, 35 worstcase iteration bound, 12
spectral norm, xvi, 232 Dikintype algorithm, 108
spectral radius, xvi, 143, 232 dual scaling method, 92
spherical simplex, 178 long step dual log-barrier method, 92
 INDEX 283

Mizuno-Todd-Ye predictorcorrector al- primaldual pathfollowing algorithm with

gorithm, 128 full NT steps, 124
short step primal log-barrier method, 88
NT potential reduction method, 146
primaldual affinescaling algorithm, 112 ZPP, 158
Applied Optimization

1. D.-Z. Du and D.F. Hsu (eds.): Combinatorial Network Theory. 1996

ISBN 0-7923-3777-8
2. M.J. Panik: Linear Programming: Mathematics, Theory and Algorithms. 1996
ISBN 0-7923-3782-4
3. R.B. Kearfott and V. Kreinovich (eds.): Applications of Interval Computations.
1996 ISBN 0-7923-3847-2
4. N. Hritonenko and Y. Yatsenko: Modeling and Optimimization of the Lifetime of
Technology. 1996 ISBN 0-7923-4014-0
5. T. Terlaky (ed.): Interior Point Methods of Mathematical Programming. 1996
ISBN 0-7923-4201-1
6. B. Jansen: Interior Point Techniques in Optimization. Complementarity, Sensitivity
and Algorithms. 1997 ISBN 0-7923-4430-8
7. A. Migdalas, P.M. Pardalos and S. Story (eds.): Parallel Computing in Optimization.
1997 ISBN 0-7923-4583-5
8. F. A. Lootsma: Fuzzy Logic for Planning and Decision Making. 1997
ISBN 0-7923-4681-5
9. J.A. dos Santos Gromicho: Quasiconvex Optimization and Location Theory. 1998
ISBN 0-7923-4694-7
10. V. Kreinovich, A. Lakeyev, J. Rohn and P. Kahl: Computational Complexity and
Feasibility of Data Processing and Interval Computations. 1998
ISBN 0-7923-4865-6
11. J. Gil-Aluja: The Interactive Management of Human Resources in Uncertainty. 1998
ISBN 0-7923-4886-9
12. C. Zopounidis and A.I. Dimitras: Multicriteria Decision Aid Methods for the Predic-
tion of Business Failure. 1998 ISBN 0-7923-4900-8
13. F. Giannessi, S. Komlsi and T. Rapcsk (eds.): New Trends in Mathematical Pro-
gramming. Homage to Steven Vajda. 1998 ISBN 0-7923-5036-7
14. Ya-xiang Yuan (ed.): Advances in Nonlinear Programming. Proceedings of the 96
International Conference on Nonlinear Programming. 1998 ISBN 0-7923-5053-7
15. W.W. Hager and P.M. Pardalos: Optimal Control. Theory, Algorithms, and Applica-
tions. 1998 ISBN 0-7923-5067-7
16. Gang Yu (ed.): Industrial Applications of Combinatorial Optimization. 1998
ISBN 0-7923-5073-1
17. D. Braha and O. Maimon (eds.): A Mathematical Theory of Design: Foundations,
Algorithms and Applications. 1998 ISBN 0-7923-5079-0
Applied Optimization

18. O. Maimon, E. Khmelnitsky and K. Kogan: Optimal Flow Control in Manufacturing.

Production Planning and Scheduling. 1998 ISBN 0-7923-5106-1
19. C. Zopounidis and P.M. Pardalos (eds.): Managing in Uncertainty: Theory and Prac-
tice. 1998 ISBN 0-7923-5110-X
20. A.S. Belenky: Operations Research in Transportation Systems: Ideas and Schemes
of Optimization Methods for Strategic Planning and Operations Management. 1998
ISBN 0-7923-5157-6
21. J. Gil-Aluja: Investment in Uncertainty. 1999 ISBN 0-7923-5296-3
22. M. Fukushima and L. Qi (eds.): Reformulation: Nonsmooth, Piecewise Smooth,
Semismooth and Smooting Methods. 1999 ISBN 0-7923-5320-X
23. M. Patriksson: Nonlinear Programming and Variational Inequality Problems. A Uni-
fied Approach. 1999 ISBN 0-7923-5455-9
24. R. De Leone, A. Murli, P.M. Pardalos and G. Toraldo (eds.): High Performance
Algorithms and Software in Nonlinear Optimization. 1999 ISBN 0-7923-5483-4
25. A. Schbel: Locating Lines and Hyperplanes. Theory and Algorithms. 1999
ISBN 0-7923-5559-8
26. R.B. Statnikov: Multicriteria Design. Optimization and Identification. 1999
ISBN 0-7923-5560-1
27. V. Tsurkov and A. Mironov: Minimax under Transportation Constrains. 1999
ISBN 0-7923-5609-8
28. V.I. Ivanov: Model Development and Optimization. 1999 ISBN 0-7923-5610-1
29. F. A. Lootsma: Multi- Criteria Decision Analysis via Ratio and Difference Judgement.
1999 ISBN 0-7923-5669-1
30. A. Eberhard, R. Hill, D. Ralph and B.M. Glover (eds.): Progress in Optimization.
Contributions from Australasia. 1999 ISBN 0-7923-5733-7
31. T. Hrlimann: Mathematical Modeling and Optimization. An Essay for the Design
of Computer-Based Modeling Tools. 1999 ISBN 0-7923-5927-5
32. J. Gil-Aluja: Elements for a Theory of Decision in Uncertainty. 1999
ISBN 0-7923-5987-9
33. H. Frenk, K. Roos, T. Terlaky and S. Zhang (eds.): High Performance Optimization.
1999 ISBN 0-7923-6013-3
34. N. Hritonenko and Y. Yatsenko: Mathematical Modeling in Economics, Ecology and
the Environment. 1999 ISBN 0-7923-6015-X
35. J. Virant: Design Considerations of Time in Fuzzy Systems. 2000
ISBN 0-7923-6100-8
Applied Optimization

36. G. Di Pillo and F. Giannessi (eds.): Nonlinear Optimization and Related Topics. 2000
ISBN 0-7923-6109-1
37. V. Tsurkov: Hierarchical Optimization and Mathematical Physics. 2000
ISBN 0-7923-6175-X
38. C. Zopounidis and M. Doumpos: Intelligent Decision Aiding Systems Based on
Multiple Criteria for Financial Engineering. 2000 ISBN 0-7923-6273-X
39. X. Yang, A.I. Mees, M. Fisher and L.Jennings (eds.): Progress in Optimization.
Contributions from Australasia. 2000 ISBN 0-7923-6286-1
40. D. Butnariu and A.N. Iusem: Totally Convex Functions for Fixed Points Computation
and Infinite Dimensional Optimization. 2000 ISBN 0-7923-6287-X
41. J. Mockus: A Set of Examples of Global and Discrete Optimization. Applications of
Bayesian Heuristic Approach. 2000 ISBN 0-7923-6359-0
42. H. Neunzert and A.H. Siddiqi: Topics in Industrial Mathematics. Case Studies and
Related Mathematical Methods. 2000 ISBN 0-7923-6417-1
43. K. Kogan and E. Khmelnitsky: Scheduling: Control-Based Theory and Polynomial-
Time Algorithms. 2000 ISBN 0-7923-6486-4
44. E. Triantaphyllou: Multi-Criteria Decision Making Methods. A Comparative Study.
2000 ISBN 0-7923-6607-7
45. S.H. Zanakis, G. Doukidis and C. Zopounidis (eds.): Decision Making: Recent Devel-
opments and Worldwide Applications. 2000 ISBN 0-7923-6621 -2
46. G.E. Stavroulakis: Inverse and Crack Identification Problems in Engineering Mech-
anics. 2000 ISBN 0-7923-6690-5
47. A. Rubinov and B. Glover (eds.): Optimization and Related Topics. 2001
ISBN 0-7923-6732-4
48. M. Pursula and J. Niittymki (eds.): Mathematical Methods on Optimization in Trans-
portation Systems. 2000 ISBN 0-7923-6774-X
49. E. Cascetta: Transportation Systems Engineering: Theory and Methods. 2001
ISBN 0-7923-6792-8
50. M.C. Ferris, O.L. Mangasarian and J.-S. Pang (eds.): Complementarity: Applications,
Algorithms and Extensions. 2001 ISBN 0-7923-6816-9
51. V. Tsurkov: Large-scale Optimization Problems and Methods. 2001
ISBN 0-7923-6817-7
52. X. Yang, K.L. Teo and L. Caccetta (eds.): Optimization Methods and Applications.
2001 ISBN 0-7923-6866-5
53. S.M. Stefanov: Separable Programming Theory and Methods. 2001
ISBN 0-7923-6882-7
Applied Optimization

54. S.P. Uryasev and P.M. Pardalos (eds.): Stochastic Optimization: Algorithms and
Applications. 2001 ISBN 0-7923-6951 -3
55. J. Gil-Aluja (ed.): Handbook of Management under Uncertainty. 2001
ISBN 0-7923-7025-2
56. B.-N. Vo, A. Cantoni and K.L. Teo: Filter Design with Time Domain Mask Con-
straints: Theory and Applications. 2001 ISBN 0-7923-7138-0
57. S. Zlobec: Stable Parametric Programming. 2001 ISBN 0-7923-7139-9
58. M.G. Nicholls, S. Clarke and B. Lehaney (eds.): Mixed-Mode Modelling: Mixing
Methodologies for Organisational Intervention. 2001 ISBN 0-7923-7151-8
59. F. Giannessi, P.M. Pardalos and T. Rapcsk (eds.): Optimization Theory. Recent
Developments from Mtrahza. 2001 ISBN 1 -4020-0009-X
60. K.M. Hangos, R. Lakner and M. Gerzson: Intelligent Control Systems. An Introduc-
tion with Examples. 2001 ISBN 1-4020-0134-7
61. D. Gstach: Estimating Output-Specific Efficiencies. 2002 ISBN 1-4020-0483-4
62. J. Geunes, P.M. Pardalos and H.E. Romeijn (eds.): Supply Chain Management:
Models, Applications, and Research Directions. 2002 ISBN 1-4020-0487-7
63. M. Gendreau and P. Marcotte (eds.): Transportation and Network Analysis: Current
Trends. Miscellanea in Honor of Michael Florian. 2002 ISBN 1-4020-0488-5
64. M. Patriksson and M. Labb (eds.): Transportation Planning. State of the Art. 2002
ISBN 1-4020-0546-6
65. E. de Klerk: Aspects of Semidefinite Programming. Interior Point Algorithms and
Selected Applications. 2002 ISBN 1-4020-0547-4

KLUWER ACADEMIC PUBLISHERS DORDRECHT / BOSTON / LONDON