A recursive least-squares algorithm for pairwise comparison matrices

Pairwise comparison matrices are commonly used for setting priorities among competing objects. In a leading decision making method called the analytic hierarchy process the principal right eigenvector components represent the weights of the alternatives. The direct least-squares method extracts the weight vector by first finding a rank-one matrix which minimizes the Euclidean distance from the original ratio matrix. We develop a recursive least-squares algorithm and reveal a striking correspondence between these two approaches for these matrices. The recursion applies for merely positive matrices also. We prove that a convergent iteration leads to matrices by which the Perron-eigenvectors and the Perron approximation of the original matrix may be produced. We show that certain useful properties of the recursion advance the development of reliable measures of perturbations of transitive matrices. Numerical analysis is included for a macroeconomic problem taken from the literature.     

Ranking by pairwise comparisons for Swiss-system tournaments

Pairwise comparison matrices are widely used in multicriteria decision making. This article applies incomplete pairwise comparison matrices in the area of sport tournaments, namely proposing alternative rankings for the 2010 Chess Olympiad Open tournament. It is shown that results are robust regarding scaling technique. In order to compare different rankings, a distance function is introduced with the aim of taking into account the subjective nature of human perception. Analysis of the weight vectors implies that methods based on pairwise comparisons have common roots. Visualization of the results is provided by multidimensional scaling on the basis of the defined distance. The proposed rankings give in some cases intuitively better outcome than currently used lexicographical orders.     

具有四个不同特征值的竞赛矩阵

设A是n阶竞赛矩阵，k是非负整数。文[3]刻划了恰好有三个不同特征值的n阶竞赛矩阵，文[4]刻划了恰好有四个不同特征值并且0作为一个一重特征值的n阶竞赛矩阵。在这篇文章中我们主要研究了两个问题：（1）讨论当k是A的特征值时A的性质。（2）刻划恰好有四个不同特征值并且k作为一个一重特征值的全部n阶竞赛矩阵。     

Tournament matrices and their generalizations,I.

If M is any complex matrix with rank (M + M ^* + I) = 1, we show that any eigenvalue of M that is not geometrically simple has 1/2 for its real part. This generalizes a recent finding of de Caen and Hoffman: the rank of any n × n tournament matrix is at least n ? 1. We extend several spectral properties of tournament matrices to this and related types of matrices. For example, we characterize the singular real matrices M with 0 diagonal for which rank (M + M^T + I) = 1 and we characterize the vectors that can be in the kernels of such matrices. We show that singular, irreducible n × n tournament matrices exist if and only n? {2,3,4,5} and exhibit many infinite families of such matrices. Connections with signed digraphs are explored and several open problems are presented.     

正则竞赛矩阵的数目和竞赛矩阵的整数特征值

本文给出了ｎ阶正则竞赛矩阵的数目的一个下界，该下界优于文献中的结果；讨论了正竞赛矩阵的性质；得到了整数１为竞赛矩阵的特征值的等价条件及这类矩阵的谱根与得分向量之间的关系．     

具有固定得分向量的竞赛矩阵的数目

本文考虑以允许平局的单循环比赛为模型的竞赛图（二重完全图）的定向图的邻接矩阵（竞赛矩阵）．给出了具有特殊得分向量的竞赛矩阵的数目,得到了具有ｎ阶强有效得分向量的竞赛矩阵的数目的下确界,并给出了达到此下界的得分向量的刻划．     

Bi-criteria evolution strategy in estimating weights from the AHP ratio-scale matrices

The problem of deriving weights from ratio-scale matrices in an analytic hierarchy process (AHP) is addressed by researchers worldwide. There are various ways to solve the problem that are generally grouped into simple matrix and optimization methods. All methods have received criticism regarding the accuracy of derived weights, and different criteria are in use to compare the weights obtained from different methods. Because the set of Pareto non-dominated solutions (weights) is unknown and for inconsistent matrices is indefinite, a bi-criterion optimization approach is proposed for manipulating such matrices. The problem-specific evolution strategy algorithm (ESA) is implemented for a robust stochastic search over a feasible indefinite solution space. The fitness function is defined as a scalar vector function composed of the common error measure, i.e. the Euclidean distance and a minimum violation error that accounts for no violation of the rank ordering. The encoding scheme and other components of the search engine are adjusted to preserve the imposed constraints related to the required normalized values of the weights. The solutions generated by the proposed approach are compared with solutions obtained by five well-known prioritization techniques for three judgment matrices taken from the literature. In these and other test applications, the prioritization method that uses the entitled weights estimation by evolution strategy algorithm (WEESA) appears to be superior to other methods if only two, the most commonly used methods, are applied: the Euclidean distance and minimum violation exclusion criteria.     

Partial and multiple match tournaments

This paper generalizes the usual tournament structure to include both partial (some pairs of players do not compete) and multiple match (some pairs of players may compete more than once) cases. Beginning with the set of all partial tournaments, a set of axioms is introduced which any distance function on this set should satisfy. In the presence of these axioms, a unique distance function (the l^p norm) is shown to exist. It is then shown that the minimum distance tournament, from the multiple match tournaments provided, is also the minimum violations ranking.     

Tournament screening cum EBIC for feature selection with high-dimensional feature spaces

The feature selection characterized by relatively small sample size and extremely high-dimensional feature space is common in many areas of contemporary statistics. The high dimensionality of the feature space causes serious difficulties: (i) the sample correlations between features become high even if the features are stochastically independent; (ii) the computation becomes intractable. These difficulties make conventional approaches either inapplicable or inefficient. The reduction of dimensionality of the feature space followed by low dimensional approaches appears the only feasible way to tackle the problem. Along this line, we develop in this article a tournament screening cum EBIC approach for feature selection with high dimensional feature space. The procedure of tournament screening mimics that of a tournament. It is shown theoretically that the tournament screening has the sure screening property, a necessary property which should be satisfied by any valid screening procedure. It is demonstrated by numerical studies that the tournament screening cum EBIC approach enjoys desirable properties such as having higher positive selection rate and lower false discovery rate than other approaches. Zehua Chen was supported by Singapore Ministry of Educations ACRF Tier 1 (Grant No. R-155-000-065-112). Jiahua Chen was supported by the National Science and Engineering Research Countil of Canada and MITACS, Canada.     

Skew-prime polynomial matrices: The polynomial-model approach

We examine the concept of skew-primeness of polynomial matrices in terms of the associated polynomial model. It is shown that skew-primeness can be characterized in terms of the property of decomposition of a vector space relative to an endomorphism. This basic result is then applied to the special case of nonsingular polynomial matrices. We investigate the nonuniqueness of skew-complements of a skew-prime pair. It is shown that the space of equivalence classes of skew-complements is in bijective correspondence with a finite-dimensional linear space. Finally, the equivalence of the solutions to the problem of output regulation with internal stability obtained via geometric methods and via polynomial matrix techniques is shown explicitly.     

Intervals of totally nonnegative matrices

Totally nonnegative matrices, i.e., matrices having all their minors nonnegative, and matrix intervals with respect to the checkerboard ordering are considered. It is proven that if the two bound matrices of such a matrix interval are nonsingular and totally nonnegative (and in addition all their zero minors are identical) then all matrices from this interval are also nonsingular and totally nonnegative (with identical zero minors).     

Asymmetric distances,semidirected networks and majority in Fermat–Weber problems

The Fermat–Weber problem is considered with distance defined by a quasimetric, an asymmetric and possibly nondefinite generalisation of a metric. In such a situation a distinction has to be made between sources and destinations. We show how the classical result of optimality at a destination or a source with majority weight, valid in a metric space, may be generalized to certain quasimetric spaces. The notion of majority has however to be strengthened to supermajority, defined by way of a measure of the asymmetry of the distance, which should be finite. This extended majority theorem applies to most asymmetric distance measures previously studied in literature, since these have finite asymmetry measure. Perhaps the most important application of quasimetrics arises in semidirected networks, which may contain edges of different (possibly zero) length according to direction, or directed edges. Distance in a semidirected network does not necessarily have finite asymmetry measure. But it is shown that an adapted majority result holds nevertheless in this important context, thanks to an extension of the classical node-optimality result to semidirected networks with constraints. Finally the majority theorem is further extended to Fermat–Weber problems with mixed asymmetric distances.     

Self-learning K-means clustering: a global optimization approach

An appropriate distance is an essential ingredient in various real-world learning tasks. Distance metric learning proposes to study a metric, which is capable of reflecting the data configuration much better in comparison with the commonly used methods. We offer an algorithm for simultaneous learning the Mahalanobis like distance and K-means clustering aiming to incorporate data rescaling and clustering so that the data separability grows iteratively in the rescaled space with its sequential clustering. At each step of the algorithm execution, a global optimization problem is resolved in order to minimize the cluster distortions resting upon the current cluster configuration. The obtained weight matrix can also be used as a cluster validation characteristic. Namely, closeness of such matrices learned during a sample process can indicate the clusters readiness; i.e. estimates the true number of clusters. Numerical experiments performed on synthetic and on real datasets verify the high reliability of the proposed method.     

The power and Arnoldi methods in an algebra of circulants

Circulant matrices play a central role in a recently proposed formulation of three‐way data computations. In this setting, a three‐way table corresponds to a matrix where each ‘scalar’ is a vector of parameters defining a circulant. This interpretation provides many generalizations of results from matrix or vector‐space algebra. These results and algorithms are closely related to standard decoupling techniques on block‐circulant matrices using the fast Fourier transform. We derive the power and Arnoldi methods in this algebra. In the course of our derivation, we define inner products, norms, and other notions. These extensions are straightforward in an algebraic sense, but the implications are dramatically different from the standard matrix case. For example, the number of eigenpairs may exceed the dimension of the matrix, although it is still polynomial in it. It is thus necessary to take an extra step and carefully select a smaller, canonical set of size equal to the dimension of the matrix from which all possible eigenpairs can be formed. Copyright © 2012 John Wiley & Sons, Ltd.     

On the eigenvalues of double band matrices

We consider matrices containing two diagonal bands of positive entries. We show that all eigenvalues of such matrices are of the form rζ, where r is a nonnegative real number and ζ is a pth root of unity, where p is the period of the matrix, which is computed from the distance between the bands. We also present a problem in the asymptotics of spectra in which such double band matrices are perturbed by banded matrices.     

A note on “Methods for constructing distance matrices and the inverse eigenvalue problem”

In this paper, a symmetric nonnegative matrix with zero diagonal and given spectrum, where exactly one of the eigenvalues is positive, is constructed. This solves the symmetric nonnegative eigenvalue problem (SNIEP) for such a spectrum. The construction is based on the idea from the paper Hayden, Reams, Wells, “Methods for constructing distance matrices and the inverse eigenvalue problem”. Some results of this paper are enhanced. The construction is applied for the solution of the inverse eigenvalue problem for Euclidean distance matrices, under some assumptions on the eigenvalues.     

The geometry of III-conditioning

Numerous problems in numerical analysis, including matrix inversion, eigen-value calculations, and polynomial zero finding, share the following property: the difficulty of solving a given problem is large when the distance from that problem to the nearest “ill-posed” one is small. For example, the closer a matrix is to the set of noninvertible matrices, the larger its condition number with respect to inversion. We show that the sets of ill-posed problems for matrix inversion, eigen-problems, and polynomial zero finding all have a common algebraic and geometric structure which lets us compute the probability distribution of the distance from a “random” problem to the set. From this probability distribution we derive, for example, the distribution of the condition number of a random matrix. We examine the relevance of this theory to the analysis and construction of numerical algorithms destined to be run in finite precision arithmetic.     

A note on symmetry reduction for circular traveling tournament problems

The traveling tournament problem (TTP) consists of finding a distance-minimal double round-robin tournament where the number of consecutive breaks is bounded. Easton et al. (2001) introduced the so-called circular TTP instances, where venues of teams are located on a circle. The distance between neighboring venues is one, so that the distance between any pair of teams is the distance on the circle. It is empirically proved that these instances are very hard to solve due to the inherent symmetry. This note presents new ideas to cut off essentially identical parts of the solution space. Enumerative solution approaches, e.g. relying on branch-and-bound, benefit from this reduction. We exemplify this benefit by modifying the DFS∗ algorithm of Uthus et al. (2009) and show that speedups can approximate factor 4n.     

Permuting matrices to avoid forbidden submatrices

This paper attaches a frame to a natural class of combinatorial problems and points out that this class includes many important special cases.

A matrix M is said to avoid a set of matrices if M does not contain any element of as (ordered) submatrix. For a fixed set of matrices, we consider the problem of deciding whether the rows and columns of a matrix can be permuted in such a way that the resulting matrix M avoids all matrices in .

We survey several known and new results on the algorithmic complexity of this problem, mostly dealing with (0,1)-matrices. Among others, we will prove that the problem is polynomial time solvable for many sets containing a single, small matrix and we will exhibit some example sets for which the problem is NP-complete.     

Conditional Gradient Method for Double-Convex Fractional Programming Matrix Problems

We consider the problem of optimizing the ratio of two convex functions over a closed and convex set in the space of matrices. This problem appears in several applications and can be classified as a double-convex fractional programming problem. In general, the objective function is nonconvex but, nevertheless, the problem has some special features. Taking advantage of these features, a conditional gradient method is proposed and analyzed, which is suitable for matrix problems. The proposed scheme is applied to two different specific problems, including the well-known trace ratio optimization problem which arises in many engineering and data processing applications. Preliminary numerical experiments are presented to illustrate the properties of the proposed scheme.