Overall performance evaluation: new bounded DEA models against unreachability of efficiency

Data envelopment analysis (DEA) performance evaluation can be implemented from either optimistic or pessimistic perspectives. For an overall performance evaluation from both perspectives, bounded DEA models are introduced to evaluate decision making units (DMUs) in terms of interval efficiencies. This paper reveals unreachability of efficiency and distortion of frontiers associated with the existing bounded DEA models. New bounded DEA models against these problems are proposed by integrating the archetypal optimistic and pessimistic DEA models into a model with bounded efficiency. It provides a new way of deriving empirical estimates of efficiency frontiers in tune with that identified by the archetypal models. Without distortion of frontiers, all DMUs reach interval efficiencies in accordance with that determined by the archetypal models. A unified evaluation and classification result is derived and the efficiency relationships between DMUs are preserved. It is shown that the newly proposed models are more reliable for overall performance evaluation in practice, as illustrated empirically by two examples.     

Data envelopment analysis,operational research and uncertainty

This paper discusses a number of applications of data envelopment analysis and the nature of uncertainty in those applications. It then reviews the key approaches to handling uncertainty in data envelopment analysis (DEA) (imprecise DEA, bootstrapping, Monte Carlo simulation and chance constrained DEA) and considers their suitability for modelling the applications. The paper concludes with suggestions about the challenges facing an operational research analyst in applying DEA in real-world situations.     

Guidelines for using variable selection techniques in data envelopment analysis

Model misspecification has significant impacts on data envelopment analysis (DEA) efficiency estimates. This paper discusses the four most widely-used approaches to guide variable specification in DEA. We analyze efficiency contribution measure (ECM), principal component analysis (PCA-DEA), a regression-based test, and bootstrapping for variable selection via Monte Carlo simulations to determine each approach’s advantages and disadvantages. For a three input, one output production process, we find that: PCA-DEA performs well with highly correlated inputs (greater than 0.8) and even for small data sets (less than 300 observations); both the regression and ECM approaches perform well under low correlation (less than 0.2) and relatively larger data sets (at least 300 observations); and bootstrapping performs relatively poorly. Bootstrapping requires hours of computational time whereas the three other methods require minutes. Based on the results, we offer guidelines for effectively choosing among the four selection methods.     

A Class of Chance Constrained Multi-objective Portfolio Selection Model Under Fuzzy Random Environment

This paper deals with a class of chance constrained portfolio selection problems in the fuzzy random decision making system. An integrated fuzzy random portfolio selection model with a chance constraint is proposed on the basis of the mean-variance model and the safety-first model. According to different definitions of chance, we consider two types of fuzzy random portfolio selection models: one is for the optimistic investors and the other is for the pessimistic investors. In order to deal with the fuzzy random models, we develop a few theorems on the variances of fuzzy random returns and the equivalent partitions of two types of chance constraints. We then transform the fuzzy random portfolio selection models into their equivalent crisp models. We further employ the ε-constraint method to obtain the efficient frontier. Finally, we apply the proposed models and approaches to the Chinese stock market as an illustration.     

A note on tight extensive game forms

Associated with each extensive game form are the α- and β-effectivity functions,E _α andE _β. These functions characterize the effective capabilities which coalitions command under pessimistic and optimistic cooperative behavior respectively. A game form is said to be tight whenE _α ≡E _β, in which case pessimistic, optimistic, and intermediate types of behavior all support the same effective power. Dalkey's condition for Nash consistency, essentially perfect information, is found to be a necessary and sufficient condition for a game form to be tight.     

Some Schroeder-Bernstein type theorems for Banach spaces

We first introduce the notion of (p,q,r)-complemented subspaces in Banach spaces, where p,q,r∈N. Then, given a couple of triples {(p,q,r),(s,t,u)} in N and putting Λ=(q+r−p)(t+u−s)−ru, we prove partially the following conjecture: For every pair of Banach spaces X and Y such that X is (p,q,r)-complemented in Y and Y is (s,t,u)-complemented in X, we have that X is isomorphic Y if and only if one of the following conditions holds:

(a)

Λ≠0, Λ divides p−q and s−t, p=1 or q=1 or s=1 or t=1.

(b)

p=q=s=t=1 and gcd(r,u)=1.

The case {(2,1,1),(2,1,1)} is the well-known Pe?czyński's decomposition method. Our result leads naturally to some generalizations of the Schroeder-Bernstein problem for Banach spaces solved by W.T. Gowers in 1996.     

On linear codes whose weights and length have a common divisor

In this paper we prove that a set of points (in a projective space over a finite field of q elements), which is incident with 0 mod r points of every hyperplane, has at least (r−1)q+(p−1)r points, where 1<r<q=ph, p prime. An immediate corollary of this theorem is that a linear code whose weights and length have a common divisor r<q and whose dual minimum distance is at least 3, has length at least (r−1)q+(p−1)r. The theorem, which is sharp in some cases, is a strong generalisation of an earlier result on the non-existence of maximal arcs in projective planes; the proof involves polynomials over finite fields, and is a streamlined and more transparent version of the earlier one.     

Enumeration of non-crossing pairings on bit strings

A non-crossing pairing on a bit string is a matching of 1s and 0s in the string with the property that the pairing diagram has no crossings. For an arbitrary bit-string w=p₁¹q₁⁰…pr¹qr⁰, let φ(w) be the number of such pairings. This enumeration problem arises when calculating moments in the theory of random matrices and free probability, and we are interested in determining useful formulas and asymptotic estimates for φ(w). Our main results include explicit formulas in the “symmetric” case where each pi=qi, as well as upper and lower bounds for φ(w) that are uniform across all words of fixed length and fixed r. In addition, we offer more refined conjectural expressions for the upper bounds. Our proofs follow from the construction of combinatorial mappings from the set of non-crossing pairings into certain generalized “Catalan” structures that include labeled trees and lattice paths.     

The interval efficiency based on the optimistic and pessimistic points of view

Data envelopment analysis (DEA) is a data-oriented approach for evaluating the performances of a set of peer entities called decision-making units (DMUs), whose performance is determined based on multiple measures. The traditional DEA, which is based on the concept of efficiency frontier (output frontier), determines the best efficiency score that can be assigned to each DMU. Based on these scores, DMUs are classified into DEA-efficient (optimistic efficient) or DEA-non-efficient (optimistic non-efficient) units, and the DEA-efficient DMUs determine the efficiency frontier. There is a comparable approach which uses the concept of inefficiency frontier (input frontier) for determining the worst relative efficiency score that can be assigned to each DMU. DMUs on the inefficiency frontier are specified as DEA-inefficient or pessimistic inefficient, and those that do not lie on the inefficient frontier, are declared to be DEA-non-inefficient or pessimistic non-inefficient. In this paper, we argue that both relative efficiencies should be considered simultaneously, and any approach that considers only one of them will be biased. For measuring the overall performance of the DMUs, we propose to integrate both efficiencies in the form of an interval, and we call the proposed DEA models for efficiency measurement the bounded DEA models. In this way, the efficiency interval provides the decision maker with all the possible values of efficiency, which reflect various perspectives. A numerical example is presented to illustrate the application of the proposed DEA models.     

Cost efficiency measurement with price uncertainty: a DEA application to bank branch assessments

This paper enhances cost efficiency measurement methods to account for different scenarios relating to input price information. These consist of situations where prices are known exactly at each decision making unit (DMU) and situations with incomplete price information. The main contribution of this paper consists of the development of a method for the estimation of upper and lower bounds for the cost efficiency (CE) measure in situations of price uncertainty, where only the maximal and minimal bounds of input prices can be estimated for each DMU. The bounds of the CE measure are obtained from assessments in the light of the most favourable price scenario (optimistic perspective) and the least favourable price scenario (pessimistic perspective). The assessments under price uncertainty are based on extensions to the Data Envelopment Analysis (DEA) model that incorporate weight restrictions of the form of input cone assurance regions. The applicability of the models developed is illustrated in the context of the analysis of bank branch performance. The results obtained in the case study showed that the DEA models can provide robust estimates of cost efficiency even in situations of price uncertainty.     

On exact order estimates of N-widths of classes of functions analytic in a simply connected domain

In the spaces E _q(Ω), 1 < q < ∞, introduced by Smirnov, we obtain exact order estimates of projective and spectral n-widths of the classes W ^r E _p(Ω) and W ^r E _p(Ω)Ф in the case where p and q are not equal. We also indicate extremal subspaces and operators for the approximative values under consideration.     

Ranking DMUs by ideal points with interval data in DEA

An original DEA model is to evaluate each DMU optimistically, but the interval DEA model proposed in this paper has been formulated to obtain an efficiency interval consisting of evaluations from both the optimistic and the pessimistic viewpoints.     

On a functional equation generalizing the class of semistable distributions

With ?(p),p≥0 the Laplace-Stieltjes transform of some infinitely divisible probability distribution, we consider the solutions to the functional equation ?(p-e ^?pβΠ _i=1 ^m ?^γi (c _i p) for somem≥1,c _i>0, γ _i >0,i=1., …,m, β ε ®. We supply its complete solutions in terms of semistable distributions (the ones obtained whenm=1). We then show how to obtain these solutions as limit laws (r → ∞) of normalized Poisson sums of iid samples when the Poisson intensity λ(r) grows geometrically withr.     

Class number one quadratic fields and solvability of some Pellian equations

Consider these two types of positive square-free integers d≠ 1 for which the class number h of the quadratic field Q(√d) is odd: (1) d is prime∈ 1(mod 8), or d=2q where q is prime ≡ 3 (mod 4), or d=qr where q and r are primes such that q≡ 3 (mod 8) and r≡ 7 (mod 8); (2) d is prime ≡ 1 (mod 8), or d=qr where q and r are primes such that q≡r≡ 3 or 7 (mod 8). For d of type (2) (resp. (1)), let Π be the set of all primes (resp. odd primes) p∈N satisfying (d/p) = 1. Also, let δ :=0 (resp. δ :=1) if d≡ 2,3 (mod 4) (resp. d≡ 1 (mod 4)). Then the following are equivalent: (a) h=1; (b) For every p∈П at least one of the two Pellian equations Z ²-dY ² = ±4^δ p is solvable in integers. (c) For every p∈П the Pellian equation W ²-dV ² = 4^δ p ² has a solution (w,v) in integers such that gcd (w,v) divides 2^δ.     

Estimates for L p -norms of simple partial fractions

We obtain estimates for L _p -norms of simple partial fractions in terms of their L _r -norms on bounded and unbounded segments of the real axis for various p > 1 and r > 1 (S. M. Nikolskii type inequalities). We adduce examples and remarks concerning sharpness of the inequalities and area of their application.     

New Partial Difference Sets in Ztp2 and a Related Problem about Galois Rings

We generalize a construction of partial difference sets (PDS) by Chen, Ray-Chaudhuri, and Xiang through a study of the Teichmüller sets of the Galois rings. Let R=GR(p², t) be the Galois ring of characteristic p² and rank t with Teichmüller set T and let π:R→R/pR be the natural homomorphism. We give a construction of PDS in R with the parameters ν=p^2t, k=r(p^t−1), λ=p^t+r²−3r, μ=r²−r, where r=lp^{t−s(p, t)}, 1≤l≤p^{s(p, t)}, and s(p, t) is the largest dimension of a GF(p)-subspace W⊂R/pR such that π⁻¹(W)∩T generates a subgroup of R of rank <t. We prove that s(p, t) is the largest dimension of a GF(p)-subspace W of GF(p^t) such that dim W^p<t, where W^p is the GF(p)-space generated by {∏^p_i=1w_i∣w_i∈W, 1≤i≤p}. We determine the values of s(p, t) completely and solve a general problem about dim_EW^r for an E-vector space W in a finite extension of a finite field E. The PDS constructed here contain the family constructed by Chen, Ray-Chaudhuri, and Xiang and have a wider range of parameters.     

Properties of Euclidean and non-Euclidean distance matrices

A distance matrix D of order n is symmetric with elements $\text{?}\text{1}\text{2}\text{d}{}_{\mathrm{ij}}{}^2$, where d_ii=0. D is Euclidean when the $\text{1}\text{2}\text{n(n?1)}$ quantities d_ij can be generated as the distances between a set of n points, X (n×p), in a Euclidean space of dimension p. The dimensionality of D is defined as the least value of p=rank(X) of any generating X; in general p+1 and p+2 are also acceptable but may include imaginary coordinates, even when D is Euclidean. Basic properties of Euclidean distance matrices are established; in particular, when ρ=rank(D) it is shown that, depending on whether e^TD^?e is not or is zero, the generating points lie in either p=ρ?1 dimensions, in which case they lie on a hypersphere, or in p=ρ?2 dimensions, in which case they do not. (The notation e is used for a vector all of whose values are one.) When D is non-Euclidean its dimensionality p=r+s will comprise r real and s imaginary columns of X, and (r, s) are invariant for all generating X of minimal rank. Higher-ranking representations can arise only from p+1=(r+1)+s or p+1=r+ (s+1) or p+2=(r+1)+(s+1), so that not only are r, s invariant, but they are both minimal for all admissible representations X.     

Prescribing the binary digits of primes

We present a new result on counting primes p < N = 2 ⁿ for which r (arbitrarily placed) digits in the binary expansion of p are specified. Compared with earlier work of Harman and Katai, the restriction on r is relaxed to r < c(n/log n)^4/7. This condition results from the estimates of Gallagher and Iwaniec on zero-free regions of L-functions with ‘powerful’ conductor.     

Boundedness of paraproduct operators on RD-spaces

Let(X,d,μ) be an RD-space with "dimension" n,namely,a space of homogeneous type in the sense of Coifman and Weiss satisfying a certain reverse doubling condition.Using the Calder'on reproducing formula,the authors hereby establish boundedness for paraproduct operators from the product of Hardy spaces H p(X) × H q(X) to the Hardy space H r(X),where p,q,r ∈(n/(n + 1),∞) satisfy 1/p + 1/q = 1/r.Certain endpoint estimates are also obtained.In view of the lack of the Fourier transform in this setting,the proofs are based on the derivation of appropriate kernel estimates.     

Effects of reactive gradient term in a multi-nonlinear parabolic problem

This paper deals with parabolic equation ut=Δu+r|∇u|−aepu subject to nonlinear boundary flux ∂u/∂η=equ, where r>1, p,q,a>0. There are two positive sources (the gradient reaction and the boundary flux) and a negative one (the absorption) in the model. It is well known that blow-up or not of solutions depends on which one dominating the model, the positive or negative sources, and furthermore on the absorption coefficient for the balance case of them. The aim of the paper is to study the influence of the reactive gradient term on the asymptotic behavior of solutions. We at first determine the critical blow-up exponent, and then obtain the blow-up rate, the blow-up set as well as the spatial blow-up profile for blow-up solutions in the one-dimensional case. It turns out that the gradient term makes a substantial contribution to the formation of blow-up if and only if r?2, where the critical r=2 is such a balance situation of the two positive sources for which the effects of the gradient reaction and the boundary source are at the same level. In addition, it is observed that the gradient term with r>2 significantly affects the blow-up rate also. In fact, the gained blow-up rates themselves contain the exponent r of the gradient term. Moreover, the blow-up rate may be discontinuous with respect to parameters included in the problem due to convection. As for the influence of gradient perturbations on spatial blow-up profiles, we only need some coefficients related to r for the profile estimates, while the exponent of the profile itself is r-independent. This seems natural for boundary blow-up solutions that the spatial profiles mainly rely on the exponent of the boundary singularity.