Algorithmic aspects of mean–variance optimization in Markov decision processes

We consider finite horizon Markov decision processes under performance measures that involve both the mean and the variance of the cumulative reward. We show that either randomized or history-based policies can improve performance. We prove that the complexity of computing a policy that maximizes the mean reward under a variance constraint is NP-hard for some cases, and strongly NP-hard for others. We finally offer pseudopolynomial exact and approximation algorithms.     

Generalizations of Several Identities

ＧｅｎｅｒａｌｉｚａｔｉｏｎｓｏｆＳｅｖｅｒａｌＩｄｅｎｔｉｔｉｅｓＺｈａｎｇＺｈｉｚｈｅｎｇ（张之正）（ＤｅｐａｒｔｍｅｎｔｏｆＭａｔｈｅｍａｔｉｃｓ，ＬｕｏｙａｎｇＴｅａｃｈｅｒｓ＇ＣｏｌｌｅｇｅＬｕｏｙａｎｇ，Ｈｅｎａｎ，４７１０２２）Ａｂｓｔ...     

Convergence of a projected gradient method with trust region for nonlinear constrained optimization†

In this paper we describe a projected gradient algorithm with trust region, introducing a nondifferentiable merit function for solving nonlinear constrained optimization problems. We show that this method is globally convergent even if conditions are weak. It is also proved that, when the strict complementarity condition holds, the proposed algorithm can be solved by an equality constrained problem, allowing locally rate of superlinear convergence.     

Several Inequalities of Matrix Traces

ＳｅｖｅｒａｌＩｎｅｑｕａｌｉｔｉｅｓｏｆＭａｔｒｉｘＴｒａｃｅｓ￥ＣｈｅｎＨｏｎｇｑｉｎｇ（ＨａｉｎａｎＮｏｒｍａｌＣｏｌｌｅｇｅ）Ａｂｓｔｒａｃｔ：Ｔｈｉｓｐａｐｅｒｈａｓｍａｄｅｔｈｅｒｅｓｕｌｔｏｆｇｅｎｅｒａｌｉｚａｔｉｏｎａｎｄｔｈｅａｎ...     

Ratio-conjugation tool and its application in linear optimization models†

The ratio-conjugation-tool (RC) is constructed systematically, and its applicability for analysis of economic models is demonstrated. One of the main results establishes a general formula for RC of aggregative function. This tool is applied to the analysis of linear economic dynamics models of Neumann–Gale type which can be solved by a dynamic programming method: for each direct problem we construct the corresponding dual RC-problem. Moreover, we consider three examples of the most important economic dynamics models and look at these now from a unified general position.     

Stability of Functional Equations in Several Variables

We prove a generalization of Hyers＇ theorem on the stability of approximately additive mapping and a generalization of Badora＇s theorem on an approximate ring homomorphism. We also obtain a more general stability theorem, which gives the stability theorems on Jordan and Lie homomorphisms. The proofs of the theorems given in this paper follow essentially the D. H. Hyers-Th. M. Rassias approach to the stability of functional equations connected with S. M. Ulam＇s problem.     

Subgradient projection algorithms for constrained nonsmooth optimization II: nonlinear constraints†

In this paper, a kind of subgradient projection algorithms is established for minimizing a locally Lipschitz continuous function subject to nonlinearly smooth constraints, which is based on the idea to get a feasible and strictly descent direction by combining the ?-subgradient projection direction that attempts to satisfy the Kuhn-Tucker conditions with one corrected direction produced by a linear programming subproblem. The algorithm avoids the zigzagging phenomenon and converges to Kuhn-Tucker points, due to using the c.d.f. maps of Polak and Mayne (1985), ?active constraints and ?adjusted rules     

Several Generalized Matrix Versions of Kantorovich Inequalities

Kantorovich inequalities are old results. In this paper we give several Kantorovich-type matrix inequalities.     

Several generalizations of Tverberg’s theorem

In a generalization of Radon’s theorem, Tverberg showed that each setS of at least (d+1) (r ? 1)+1 points inR ^d has anr-partition into (pair wise disjoint) subsetsS =S ₁ ∪ … ∪S _r so that \(\bigcap\nolimits_i^r {\underline{\underline {}} } _1 \) convS _i # Ø. This note considers the following more general problems: (1) How large mustS σR ^d be to assure thatS has anr-partitionS=S ₁∪ … ∪S _r so that eachn members of the family {convS _i ～ _i-1 ^r have non-empty intersection, where 1<=n<=r. (2) How large mustS ∪R ^d be to assure thatS has anr-partition for which \(\bigcap\nolimits_i^r {\underline{\underline {}} } _1 \) convS _r is at least 1-dimensional.     

Several Results on Systems of Residue Classes

Let (m,n) and a(n) denote the g.c.d, of m, n and the residue class {x∈Z∶x≡α (mod n)} respectively. Any period of the characteristic function ofkU a_i(n_i) is called a covering period of {a_i(n_i)}_(i-1)~k.i-ITheorem Let A = {a_i(n_i)}_(i-1)~k. be a disjoint system (i. e. a_I(n_I,...,a_k(n_k) are pairwise disjoint). Let [n_I,...,n_k] (the I.c.m. of n_1,...,n_k) have the prime faetorization [n_1,...,n_k] = Πp_i~ai and T = Πp_iβi(β_i≥0 be the smallest positive covering period of A. Then     

Dynamic models of segregation†

Some segregation results from the practices of organizations, some from specialized communication systems, some from correlation with a variable that is non‐random; and some results from the interplay of individual choices. This is an abstract study of the interactive dynamics of discriminatory individual choices. One model is a simulation in which individual members of two recognizable groups distribute themselves in neighborhoods defined by reference to their own locations. A second model is analytic and deals with compartmented space. A final section applies the analytics to ‘neighborhood tipping.’ The systemic effects are found to be overwhelming: there is no simple correspondence of individual incentive to collective results. Exaggerated separation and patterning result from the dynamics of movement. Inferences about individual motives can usually not be drawn from aggregate patterns. Some unexpected phenomena, like density and vacancy, are generated. A general theory of ‘tipping’ begins to emerge.     

Genus Distributions for Several Types of Ladder-class Graphs

Calculating the genus distributions of ladder graphs is a concerned topic in topological graph theory.In this paper,we formulate several ladder-class graphs by using a starting graph iterative amalgamation with copies of a path to construct a base graph and then adding some edges to the appointed root-vertices of the base graph.By means of transfer matrix and a finer partition of the embeddings,the explicit formulas for the genus distribution polynomials of four types of ladder-class graphs are derived.     

λ-Divergence of the Fourier Series of Continuous Functions of Several Variables

In this paper, we consider the behavior of rectangular partial sums of the Fourier series of continuous functions of several variables with respect to the trigonometric system. The Fourier series is called -convergent if the limit of rectangular partial sums over all indices for which for all j and k exists. In the space of arbitrary even dimension 2m we construct an example of a continuous function with an estimate of the modulus of continuity such that its Fourier series is -divergent everywhere for any .     

Parameters optimization of selected casting processes using teaching–learning-based optimization algorithm

In the present work, mathematical models of three important casting processes are considered namely squeeze casting, continuous casting and die casting for the parameters optimization of respective processes. A recently developed advanced optimization algorithm named as teaching–learning-based optimization (TLBO) is used for the parameters optimization of these casting processes. Each process is described with a suitable example which involves respective process parameters. The mathematical model related to the squeeze casting is a multi-objective problem whereas the model related to the continuous casting is multi-objective multi-constrained problem and the problem related to the die casting is a single objective problem. The mathematical models which are considered in the present work were previously attempted by genetic algorithm and simulated annealing algorithms. However, attempt is made in the present work to minimize the computational efforts using the TLBO algorithm. Considerable improvements in results are obtained in all the cases and it is believed that a global optimum solution is achieved in the case of die casting process.     

The structure of irreducible matrices †

Let A be an irreducible matrix with index of imprimitivity h is shown that there exists a permutation matrix P such that PAP^t is in a superdiagonal block form with k nonzero blocks if and only if k divides h It is also shown that a matrix in a superdiagonal block form without zero rows or columns is irreducible if and only if the product of the superdiagonal nonzero blocks is irreducible.     

Subpermanents of doubly stochastic matrices †

It is shown that if all subpermaneats of order k of an n × n doubly stochastic matrix are equal for some k ≤ n ? 2, then all the entries of the matrix must be equal to 1/n.     

Several identities for certain products of theta functions

By means of a technique used by Carlitz and Subbarao to prove the quintuple product identity (Proc. Am. Math. Soc. 32(1):42–44, 1972), we recover a general identity (Chu and Yan, Electron. J. Comb. 14:#N7, 2007) for expanding the product of two Jacobi triple products. For applications, we briefly explore identities for certain products of theta functions φ(q), ψ(q) and modular relations for the Göllnitz-Gordon functions.     

The Proofs of Several Theorems about the Order Relation

In this note,we use the fundamental rulesof the mathematical logic to pro-ve several theorems concerned with the order relation in decision theory,in orderto demonstrate that this kind of method seems simpler,clearer,and more strict,and efficient to decrease mistakes.In the maentime,this process laeds to findinga contradition and a flaw in the previous statements of the theorems.