Validation and verification of social processes within agent-based computational organization models

The use of simulation modeling in computational analysis of organizations is becoming a prominent approach in social science research. However, relying on simulations to gain intuition about social phenomena has significant implications. While simulations may give rise to interesting macro-level phenomena, and sometimes even mimic empirical data, the underlying micro and macro level processes may be far from realistic. Yet, this realism may be important to infer results that are relevant to existing theories of social systems and to policy making. Therefore, it is important to assess not only predictive capability but also explanation accuracy of formal models in terms of the degree of realism reflected by the embedded processes. This paper presents a process-centric perspective for the validation and verification (V&V) of agent-based computational organization models. Following an overview of the role of V&V within the life cycle of a simulation study, emergent issues in agent-based organization model V&V are outlined. The notion of social contract that facilitates capturing micro level processes among agents is introduced to enable reasoning about the integrity and consistency of agent-based organization designs. Social contracts are shown to enable modular compositional verification of interaction dynamics among peer agents. Two types of consistency are introduced: horizontal and vertical consistency. It is argued that such local consistency analysis is necessary, but insufficient to validate emergent macro processes within multi-agent organizations. As such, new formal validation metrics are introduced to substantiate the operational validity of emergent macro-level behavior. Levent Yilmaz is Assistant Professor of Computer Science and Engineering in the College of Engineering at Auburn University and co-founder of the Auburn Modeling and Simulation Laboratory of the M&SNet. Dr. Yilmaz received his Ph.D. and M.S. degrees from Virginia Polytechnic Institute and State University (Virginia Tech). His research interests are on advancing the theory and methodology of simulation modeling, agent-directed simulation (to explore dynamics of socio-technical systems, organizations, and human/team behavior), and education in simulation modeling. Dr. Yilmaz is a member of ACM, IEEE Computer Society, Society for Computer Simulation International, and Upsilon Pi Epsilon. URL: http://www.eng.auburn.edu/~yilmaz     

Unique ergodicity of flows on the torus and the rotation number

We prove in this paper that any flow on the 2- torus with no singular points and periodic orbits which is generated by a vector fieldV=(P, Q) satisfyingV∈C ¹ orV∈C ⁰ andP≠0 is uniquely ergodic. Then we give an expression of the rotation number by using an invariant measure of a flow.     

Valid inequalities and restrictions for stochastic programming problems with first order stochastic dominance constraints

Stochastic dominance relations are well studied in statistics, decision theory and economics. Recently, there has been significant interest in introducing dominance relations into stochastic optimization problems as constraints. In the discrete case, stochastic optimization models involving second order stochastic dominance constraints can be solved by linear programming. However, problems involving first order stochastic dominance constraints are potentially hard due to the non-convexity of the associated feasible regions. In this paper we consider a mixed 0–1 linear programming formulation of a discrete first order constrained optimization model and present a relaxation based on second order constraints. We derive some valid inequalities and restrictions by employing the probabilistic structure of the problem. We also generate cuts that are valid inequalities for the disjunctive relaxations arising from the underlying combinatorial structure of the problem by applying the lift-and-project procedure. We describe three heuristic algorithms to construct feasible solutions, based on conditional second order constraints, variable fixing, and conditional value at risk. Finally, we present numerical results for several instances of a real world portfolio optimization problem. This research was supported by the NSF awards DMS-0603728 and DMI-0354678.     

Evaluating the quality of online optimization algorithms by discrete event simulation

A key feature of dynamic problems which offer degrees of freedom to the decision maker is the necessity for a goal-oriented decision making routine which is employed every time the logic of the system requires a decision. In this paper, we look at optimization procedures which appear as subroutines in dynamic problems and show how discrete event simulation can be used to assess the quality of algorithms: after establishing a general link between online optimization and discrete event systems, we address performance measurement in dynamic settings and derive a corresponding tool kit. We then analyze several control strategies using the methodologies discussed previously in two real world examples of discrete event simulation models: a manual order picking system and a pickup and delivery service.     

A MIP flow model for crop-rotation planning in a context of forest sustainable development

We propose a Mixed-Integer Linear Programming model for a class of multi-period crop rotation optimization problems with demand constraints and incompatibility constraints between cultivation and fallow state on a land plot. This model is applied to a case study on Madagascan farms in the scope of a sustainable development campain against deforestation, where the objective is to better control agricultural space while covering seasonal needs of farmer. We propose an efficient upper bound computation and study the variation of the minimum number of plots and total space needed in function of the unitary surface area of a plot. Numerical results associated with the Madagascan case are reported.     

A Crop Planning Problem with Fuzzy Random Profit Coefficients

The main purpose of this paper is to present a crop planning problem for agricultural management under uncertainty. It is significant that agricultural managers assign their limited farmlands to cultivation of which crops in a season. This planning is called the crop planning problem and influences their incomes for the season. Usually, the crop planning problem is formulated as a linear programming problem. But there are many uncertain factors in agricultural problems, so future profits for crops are not certain values. A linear programming model with constant profit coefficients may not reflect the environment of decision making properly. Therefore, we propose a model of crop planning with fuzzy profit coefficients, and an effective solution procedure for the model. Furthermore, we extend this fuzzy model, setting the profit coefficients as discrete randomized fuzzy numbers. We show concrete optimal solutions for each models.     

Why fix it when it ain't broken? A reply to Henning's critique

This rejoinder discusses Henning's arguments against my model for exchange systems with restricted market access. It also presents an alternative rational choice model that combines Henning's central demands with my original modeling idea (according to which restricted access has effects before the optimization). This alternative approach yields similar conclusions as my earlier analysis. A comparison with Henning's results (which refer to a scenario in which access restrictions explicitly enter the optimization program) suggests that the respective model implications depend crucially on the decision where and how restricted access is introduced into a model. From this perspective, Henning's negative assessment of my work is questionable because, in the absence of empirical evidence, he has no better rationale for modeling restricted access than I do.     

Decision with Dempster-Shafer belief functions: Decision under ignorance and sequential consistency

This paper examines proposals for decision making with Dempster-Shafer belief functions from the perspectives of requirements for rational decision under ignorance and sequential consistency. The focus is on the proposals by Jaffray & Wakker and Giang & Shenoy applied for partially consonant belief functions. We formalize the concept of sequential consistency of an evaluation model and prove results about sequential consistency of Jaffray-Wakker’s model and Giang-Shenoy’s model under various conditions. We demonstrate that the often neglected assumption about two-stage resolution of uncertainty used in Jaffray-Wakker’s model actually disambiguates the foci of a belief function, and therefore, makes it a partially consonant on the extended state space.     

Assessing the relative efficiency of decision making units using DEA models with weight restrictions

In models of data envelopment analysis (DEA), an optimal set of input and output weights is generally assumed to represent the assessed decision making unit (DMU) in the best light in comparison to all the other DMUs. The paper shows that this may not be correct if absolute weight bounds or some other weight restrictions are added to the model. A consequence may be that the model will underestimate the relative efficiency of DMUs. The incorporation of weight restrictions in a maximin DEA model is suggested. This model can be further converted to more operational forms, which are similar to the classical DEA models.     

CMARS and GAM & CQP—Modern optimization methods applied to international credit default prediction

In this paper, we apply newly developed methods called GAM & CQP and CMARS for country defaults. These are techniques refined by us using Conic Quadratic Programming. Moreover, we compare these new methods with common and regularly used classification tools, applied on 33 emerging markets’ data in the period of 1980-2005. We conclude that GAM & CQP and CMARS provide an efficient alternative in predictions. The aim of this study is to develop a model for predicting the countries’ default possibilities with the help of modern techniques of continuous optimization, especially conic quadratic programming. We want to show that the continuous optimization techniques used in data mining are also very successful in financial theory and application. By this paper we contribute to further benefits from model-based methods of applied mathematics in the financial sector. Herewith, we aim to help build up our nations.     

A finite horizon model for repairable systems with repair restrictions

In this paper, we will present a new finite horizon repair/replacement decision model and derive the structure of the optimal policy for components that have a failure intensity that is a non-decreasing function of the number of times the component has been repaired, and independent of the component's age. Furthermore, the component has physical restrictions on the number of times it can be repaired, after which the only feasible decision is to replace the component. The fundamentals of this new decision model are based on the outcomes of several case studies done by the authors. Besides presenting the model and showing the structure of the optimal policy, the model will be applied to a real industry data set, and its results discussed.     

Congestion analysis in DEA inputs under weight restrictions

Data envelopment analysis (DEA), which is used to determine the efficiency of a decision-making unit (DMU), is able to recognize the amount of input congestion. Moreover, the relative importance of inputs and outputs can be incorporated into DEA models by weight restrictions. These restrictions or a priori weights are introduced by the decision maker and lead to changes in models and efficiency interpretation. In this paper, we present an approach to determine the value of congestion in inputs under the weight restrictions. Some discussions show how weight restrictions can affect the congestion amount.     

系统数值优化方法的研究与应用

大系统具有结构复杂、功能综合、因素众多等特点,相应的规划和决策问题的求解是非常困难的。我们针对大系统工程规划和决策的特点,结合运筹学的原理,提出了大系统数值优化概念,建立了优化模型,给出了优化问题的求解方法;并选择了中石油的DS-1油田作为应用实例,进行了油田采油工程数值优化,取得了很好的效果。为一般的复杂工程系统的规划和决策提供了一个新方法。     

Cost/risk balanced management of scarce resources using stochastic programming

We consider the situation when a scarce renewable resource should be periodically distributed between different users by a Resource Management Authority (RMA). The replenishment of this resource as well as users demand is subject to considerable uncertainty. We develop cost optimization and risk management models that can assist the RMA in its decision about striking the balance between the level of target delivery to the users and the level of risk that this delivery will not be met. These models are based on utilization and further development of the general methodology of stochastic programming for scenario optimization, taking into account appropriate risk management approaches. By a scenario optimization model we obtain a target barycentric value with respect to selected decision variables. A successive reoptimization of deterministic model for the worst case scenarios allows the reduction of the risk of negative consequences derived from unmet resources demand. Our reference case study is the distribution of scarce water resources. We show results of some numerical experiments in real physical systems.     

Improved customer choice predictions using ensemble methods

In this paper various ensemble learning methods from machine learning and statistics are considered and applied to the customer choice modeling problem. The application of ensemble learning usually improves the prediction quality of flexible models like decision trees and thus leads to improved predictions. We give experimental results for two real-life marketing datasets using decision trees, ensemble versions of decision trees and the logistic regression model, which is a standard approach for this problem. The ensemble models are found to improve upon individual decision trees and outperform logistic regression.     

Avoiding infeasibility in DEA models with weight restrictions

The flexibility of weights assigned to inputs and outputs is a key aspect of DEA modeling. However, excessive weight variability and implausible weight values have led to the development of DEA models that incorporate weight restrictions, reflecting expert judgment. This in turn has created problems of infeasibility of the corresponding linear programs. We provide an existence theorem that establishes feasibility conditions for DEA multiplier programs with weight restrictions. We then propose a linear model that tests for feasibility and a nonlinear model that provides minimally acceptable adjustments to the original restrictions that render the program feasible. The analysis can be applied to restrictions on weight ratios, or to restrictions on virtual inputs or outputs.     

Multi-period stochastic portfolio optimization: Block-separable decomposition

We consider a multiperiod stochastic programming recourse model for stock portfolio optimization. The presence of various risk and policy constraints leads to significant period-by-period linkage in the model. Furthermore, the dimensionality of the model is large due to many securities under consideration. We propose exploiting block separable recourse structure as well as methods of inducing such structure within nested L-shaped decomposition. We test the model and solution methodology with a base consisting of the Standard & Poor 100 stocks and experiment with several variants of the block separable technique. These are then compared to the standard nested period-by-period decomposition algorithm. It turns out that for financial optimization models of the kind that are discussed in this paper, significant computational efficiencies can be gained with the proposed methodology.     

Toward Best Approximation of Nonlinear Systems: A Case of Models with Memory

The problem of nonlinear dynamical system modeling, considered in this paper, is motivated by restrictions arising in real-world tasks. The restrictions are that first, a system input cannot be entirely observed for one trial. Second, the system model must be subjected to the causality principle. Third, the input is corrupted by noise so that no relationship between the reference input and noise is known. Fourth, the model should have some degrees of freedom so that the associated accuracy can be regulated by a variation of these freedom degrees. We propose and justify new procedures for the nonlinear system modeling that are initialized by these motivations. The models are nonlinear and given by so called r-degree operators that can be reduced to a matrix form presentation. To satisfy the restrictions above, the matrices have special structures that we call the lower p-band matrices. The degree r of the models is the required degree of freedom. The rigorous analysis of errors associated with the presented techniques is given. Numerical experiments with real data demonstrate the efficiency of the proposed approach.