A maximum entropy approach to the newsvendor problem with partial information

In this paper, we consider the newsvendor model under partial information, i.e., where the demand distribution D is partly unknown. We focus on the classical case where the retailer only knows the expectation and variance of D. The standard approach is then to determine the order quantity using conservative rules such as minimax regret or Scarf’s rule. We compute instead the most likely demand distribution in the sense of maximum entropy. We then compare the performance of the maximum entropy approach with minimax regret and Scarf’s rule on large samples of randomly drawn demand distributions. We show that the average performance of the maximum entropy approach is considerably better than either alternative, and more surprisingly, that it is in most cases a better hedge against bad results.     

Hedging of unit-linked life insurance contracts with unobservable mortality hazard rate via local risk-minimization

In this paper we investigate the local risk-minimization approach for a combined financial-insurance model where there are restrictions on the information available to the insurance company. In particular we assume that, at any time, the insurance company may observe the number of deaths from a specific portfolio of insured individuals but not the mortality hazard rate. We consider a financial market driven by a general semimartingale and we aim to hedge unit-linked life insurance contracts via the local risk-minimization approach under partial information. The Föllmer–Schweizer decomposition of the insurance claim and explicit formulas for the optimal strategy for pure endowment and term insurance contracts are provided in terms of the projection of the survival process on the information flow. Moreover, in a Markovian framework, this leads to a filtering problem with point process observations.     

Families of relations modelling preferences under incomplete information

Let us consider a preferential information of type preference–indifference–incomparability (P, I, J), with additional information about differences in attractiveness between pairs of alternatives. The present paper offers a theoretical framework for the study of the “level of constraint” of this kind of partial preferential information. It suggests a number of structures as potential models being less demanding than the classical one in which differences in utilities can be used to represent the comparison of differences in attractiveness. The models are characterized in the more general context of families of non-complete preference structures, according to two different perspectives (called “semantico-numerical” and “matrix”). Both perspectives open the door to further practical applications connected with elicitation of the preferences of a decision maker.     

Partial information systems and the Smyth powerdomain

The dual of the join semilattice of proper compact Scott open subsets of a domain D is called the Smyth powerdomain of D. The Smyth powerdomain is used in programming semantics as a model for demonic nondeterminism. In this paper, we introduce the concept of partial information systems; and, as an application, show that the Smyth powerdomain of any domain can be realized in terms of the sub partial information systems of the domain??s corresponding information system.     

Equilibrium and optimal strategies to join a queue with partial information on service times

In this paper, we study customer equilibrium as well as socially optimal strategies to join a queue with only partial information on the service time distribution such as moments and the range. Based on such partial information, customers adopt the entropy-maximization principle to obtain the expectation of their waiting cost and decide to join or balk. We find that more information encourages customers to join the queue. And it is beneficial for decision makers to convey partial information to customers in welfare maximization but reveal full information in profit maximization.     

Deriving weights in multiple-criteria decision making with support vector machines

A key problem in Multiple-Criteria Decision Making is how to measure the importance of the different criteria when just a partial preference relation among actions is given. In this note we address the problem of constructing a linear score function (and thus how to associate weights of importance to the criteria) when a binary relation comparing actions and partial information (relative importance) on the criteria are given. It is shown that these tasks can be done viaSupport Vector Machines, an increasingly popular Data Mining technique, which reduces the search of the weights to the resolution of (a series of) nonlinear convex optimization problems with linear constraints. An interactive method is then presented and illustrated by solving a multiple-objective 0–1 knapsack problem. Extensions to the case in which data are imprecise (given by intervals) or intransitivities in strict preferences exist are outlined.     

Rectifying partial algebras over operads of complexes

Kriz and May (1995) [2] introduced partial algebras over an operad. In this paper we prove that, in the category of chain complexes, partial algebras can be functorially replaced by quasi-isomorphic algebras. In particular, partial algebras contain all of the important homological and homotopical information that genuine algebras do. Applying this result to McClure's partial algebra in McClure (2006) [5] shows that the chains of a PL-manifold are quasi-isomorphic to an E_∞-algebra.     

Information and preference reversals in lotteries

Several approaches have been proposed for evaluating information in expected utility theory. Among the most popular approaches are the expected utility increase, the selling price and the buying price. While the expected utility increase and the selling price always agree in ranking information alternatives, Hazen and Sounderpandian [11] have demonstrated that the buying price may not always agree with the other two. That is, in some cases, where the expected utility increase would value information A more highly than information B, the buying price may reverse these preferences. In this paper, we discuss the conditions under which all these approaches agree in a generic decision environment where the decision maker may choose to acquire arbitrary information bundles.     

Top-Down Fuzzy Decision Making with Partial Preference Information

This paper proposes a multi-stage decision procedure to cope with a hierarchical multiple objective decision environment in which the upper-level DM only provides partial preference information and the lower-level DM is fuzzy about the tradeoff questions such that to achieve substantially more than or equal to some values is delivered to maximize the objectives. Therefore, the procedure consists of two levels, a upper-level and a lower-level. The main idea is that after the upper-level provides partial preference information to the lower-level as a guideline of decision, the lower-level DM determines a satisfactory solution from the reduced non-dominated set in the framework of multi-objective fuzzy programs.     

Inventory control with product returns: The impact of imperfect information

Product returns are characterized by considerable uncertainty on time and quantity. In the literature on inventory management for product return environments best forecasts of future returns are associated with methods that use the most information regarding product return history. In practice, however, data is often scarce and unreliable, while forecasts based on historical data, reliable or not, are never perfect. In this paper we therefore investigate the impact of imperfect information with respect to the return process on inventory management performance. We show that in the case of imperfect information the most informed method does not necessarily lead to best performance. The results have relevant implications regarding investments in product return information systems.     

一类部分信息的随机控制问题的极值原理

在本文中,我们证明了一类部分信息的随机控制问题的极值原理的一个充分条件和一个必要条件.其中,随机控制问题的控制系统是一个由鞅和Brown运动趋动的随机偏微分方程.     

一个投资问题的信息结构

本文研究一个投资问题中的信息价值，其中信息可以是不完美的，信息价值依赖于其结构．信息结构由状态变量与信息变量联合概率(或条件概率)矩阵来描述．我们定义了矩阵的一些偏序，通过这些偏序给出了信息价值的比较关系．由这些结果，我们导出了信息价值的一些基本性质．     

Complexity of differential and integral equations

We are interested in the intrinsic difficulty (or complexity) of computing an approximate solution of the linear operator equation Lu = f. Practical examples of such problems include the cases where L is a known partial differential or integral operator. Problems of the form Lu = f are typically solved under the constraint that only partial information about f is available, such as the values of a finite number of inner products, or the values of f at a finite number of points. It is of interest to determine when algorithms which are in wide use are optimal algorithms, i.e., algorithms which produce an approximation with minimal cost. We are especially interested in determining conditions which are necessary and sufficient for the finite element method (FEM) to be optimal. For the cases of elliptic partial differential equations and of Fredholm integral equations of the second kind, we describe such a condition, in the form of an inequality involving the order of the problem and the degree of the finite element subspace. Suppose this inequality is violated; is the nonoptimality of the FEM inherent in the information used by the FEM, or is it because the FEM uses this information in a nonoptimal manner? The latter is the case; there always exists an algorithm using this information which is optimal. We also discuss the situation in which the information used by the finite element method (which consists of inner products) is not available. Suppose that the only admissible information about f consists of evaluations of f. In the case of the Fredholm problem of the second kind, this information is optimal; moreover, a finite element method in which the inner products are approximated by quadrature rules is an optimal algorithm. However there exist elliptic problems of positive order for which this new information is nonoptimal.     

Undiscounted Markov decision chains with partial information; an algorithm for computing a locally optimal periodic policy

In this paper we construct an algorithm of successive approximation type that computes a locally optimal periodic policy for a Markov decision chain with partial state information. The algorithm is applied to a queueing network with server control.The research of this author has been supported by the Netherlands Organization for Scientific Research (N.W.O.).     

The Information Content of Data in Defined Choice-Situations

This paper is concerned with what people's decisions imply about the usefulness of the data they receive. It does not deal with what might improve decision-making but with an attempt to understand what the measurement of information for decision-making involves. Operational definitions and measures of the information content of data are discussed. Ackoff and Emery's treatment of information as dependent on the choice-situation is followed, but a new concept, that of information inherent in the structure of the choice-situation is added. It is shown that the Shannon measure of uncertainty can be introduced into the Ackoff-Emery behavioural framework for dealing with measures of information. The greater clarity stemming from this partial unification of two apparently different approaches may help with the still formidable problems that are outlined.     

The stability of the entropy of degree alpha

In this paper, we first prove that the generalized fundamental equation of information depending on a positive real parameter α, is stable in the sense of Hyers and Ulam provided that α≠1, then we apply this result to prove the stability of a system of functional equations that characterizes the entropy of degree alpha or Havrda-Charvát entropy which has recently often been called the Tsallis entropy.     

A Maximum Principle for Stochastic Control with Partial Information

Abstract

We study the problem of optimal control of a jump diffusion, that is, a process which is the solution of a stochastic differential equation driven by Lévy processes. It is required that the control process is adapted to a given subfiltration of the filtration generated by the underlying Lévy processes. We prove two maximum principles (one sufficient and one necessary) for this type of partial information control. The results are applied to a partial information mean-variance portfolio selection problem in finance.     

Time-consistent investment strategy under partial information

This paper considers a mean–variance portfolio selection problem under partial information, that is, the investor can observe the risky asset price with random drift which is not directly observable in financial markets. Since the dynamic mean–variance portfolio selection problem is time inconsistent, to seek the time-consistent investment strategy, the optimization problem is formulated and tackled in a game theoretic framework. Closed-form expressions of the equilibrium investment strategy and the corresponding equilibrium value function under partial information are derived by solving an extended Hamilton–Jacobi–Bellman system of equations. In addition, the results are also given under complete information, which are need for the partial information case. Furthermore, some numerical examples are presented to illustrate the derived equilibrium investment strategies and numerical sensitivity analysis is provided.     

Anisotropic mesh refinement for finite element methods based on error reduction

In this paper, we propose an anisotropic adaptive refinement algorithm based on the finite element methods for the numerical solution of partial differential equations. In 2-D, for a given triangular grid and finite element approximating space V, we obtain information on location and direction of refinement by estimating the reduction of the error if a single degree of freedom is added to V. For our model problem the algorithm fits highly stretched triangles along an interior layer, reducing the number of degrees of freedom that a standard h-type isotropic refinement algorithm would use.     

Observable graphs

An edge-colored directed graph is observable if an agent that moves along its edges from node to node is able to determine his position in the graph after a sufficiently long observation of the edge colors, and without accessing any information about the traversed nodes. When the agent is able to determine his position only from time to time, the graph is said to be partly observable. Observability in graphs is desirable in situations where autonomous agents are moving on a network and they want to localize themselves with limited information. In this paper, we completely characterize observable and partly observable graphs and show how these concepts relate to other concepts in the literature. Based on these characterizations, we provide polynomial time algorithms to decide observability, to decide partial observability, and to compute the minimal number of observations necessary for finding the position of an agent. In particular we prove that in the worst case this minimal number of observations increases quadratically with the number of nodes in the graph. We then consider the more difficult question of assigning colors to a graph so as to make it observable and we prove that two different versions of this problem are NP-complete.