Epiconvergence of relaxed stochastic optimization problems

We consider relaxation of almost sure constraint in dynamic stochastic optimization problems and their convergence. We show an epiconvergence result relying on the Kudo convergence of $\sigma$-algebras and continuity of the objective and constraint operators. We present classical constraints and objective functions with conditions ensuring their continuity. We are motivated by a Lagrangian decomposition algorithm, known as Dual Approximate Dynamic Programming, that relies on relaxation, and can also be understood as a decision rule approach in the dual.     

Risk-averse stochastic optimal control: An efficiently computable statistical upper bound

In this paper, we discuss an application of the Stochastic Dual Dynamic Programming (SDDP) type algorithm to nested risk-averse formulations of Stochastic Optimal Control (SOC) problems. We propose a construction of a statistical upper bound for the optimal value of risk-averse SOC problems. This outlines an approach to a solution of a long standing problem in that area of research. The bound holds for a large class of convex and monotone conditional risk mappings. Finally, we show the validity of the statistical upper bound to solve a real-life stochastic hydro-thermal planning problem.     

交叉数学规划问题

本文提出了一个新的数学规划概念──交叉数学规划问题．该问题的提出是以经济问题为其背景的．许多已有的规划问题上。对偶规划问题、双水平规划问题、多目标规划问题、参数规划问题以及对策问题均可作为交叉规划问题的特例．本文除系统地给出交及数学规划问题的基本定义外，还分别对各类交叉规划问题的有关理论及求解方法进行了初步的探讨．     

The Linear Program approach in multi-chain Markov Decision Processes revisited

Linear Programming is known to be an important and useful tool for solving Markov Decision Processes (MDP). Its derivation relies on the Dynamic Programming approach, which also serves to solve MDP. However, for Markov Decision Processes with several constraints the only available methods are based on Linear Programs. The aim of this paper is to investigate some aspects of such Linear Programs, related to multi-chain MDPs. We first present a stochastic interpretation of the decision variables that appear in the Linear Programs available in the literature. We then show for the multi-constrained Markov Decision Process that the Linear Program suggested in [9] can be obtained from an equivalent unconstrained Lagrange formulation of the control problem. This shows the connection between the Linear Program approach and the Lagrange approach, that was previously used only for the case of a single constraint [3, 14, 15].     

A re-scaled twin augmented Lagrangian algorithm for saddle point seeking

In [A. Ouorou, A primal-dual algorithm for monotropic programming and its application to network optimization, Computational Optimization and Application 15 (2002) 125–143], a block-wise Gauss–Seidel method has been developed for monotropic programming problems, using two different quadratic augmented Lagrangian functions defined for the primal and the dual problems. In this paper, we extend the concept by introducing a nonlinear re-scaling principle obtained recently by Polyak [R. Polyak, Nonlinear rescaling vs smoothing technique in constrained optimization, Mathematical Programming 92 (2002) 197–235].     

Analysis of stochastic dual dynamic programming method

In this paper we discuss statistical properties and convergence of the Stochastic Dual Dynamic Programming (SDDP) method applied to multistage linear stochastic programming problems. We assume that the underline data process is stagewise independent and consider the framework where at first a random sample from the original (true) distribution is generated and consequently the SDDP algorithm is applied to the constructed Sample Average Approximation (SAA) problem. Then we proceed to analysis of the SDDP solutions of the SAA problem and their relations to solutions of the “true” problem. Finally we discuss an extension of the SDDP method to a risk averse formulation of multistage stochastic programs. We argue that the computational complexity of the corresponding SDDP algorithm is almost the same as in the risk neutral case.     

Possibilistic sequential decision making

When the information about uncertainty cannot be quantified in a simple, probabilistic way, the topic of possibilistic decision theory is often a natural one to consider. The development of possibilistic decision theory has lead to the proposition a series of possibilistic criteria, namely: optimistic and pessimistic possibilistic qualitative criteria [7], possibilistic likely dominance [2], [9], binary possibilistic utility [11] and possibilistic Choquet integrals [24]. This paper focuses on sequential decision making in possibilistic decision trees. It proposes a theoretical study on the complexity of the problem of finding an optimal strategy depending on the monotonicity property of the optimization criteria – when the criterion is transitive, this property indeed allows a polytime solving of the problem by Dynamic Programming. We show that most possibilistic decision criteria, but possibilistic Choquet integrals, satisfy monotonicity and that the corresponding optimization problems can be solved in polynomial time by Dynamic Programming. Concerning the possibilistic likely dominance criteria which is quasi-transitive but not fully transitive, we propose an extended version of Dynamic Programming which remains polynomial in the size of the decision tree. We also show that for the particular case of possibilistic Choquet integrals, the problem of finding an optimal strategy is NP-hard. It can be solved by a Branch and Bound algorithm. Experiments show that even not necessarily optimal, the strategies built by Dynamic Programming are generally very good.     

A novel approach to Bilevel nonlinear programming

Recently developed methods of monotonic optimization have been applied successfully for studying a wide class of nonconvex optimization problems, that includes, among others, generalized polynomial programming, generalized multiplicative and fractional programming, discrete programming, optimization over the efficient set, complementarity problems. In the present paper the monotonic approach is extended to the General Bilevel Programming GBP Problem. It is shown that (GBP) can be transformed into a monotonic optimization problem which can then be solved by “polyblock” approximation or, more efficiently, by a branch-reduce-and-bound method using monotonicity cuts. The method is particularly suitable for Bilevel Convex Programming and Bilevel Linear Programming.      

Application of general semi-infinite programming to lapidary cutting problems

We consider a volume maximization problem arising in gemstone cutting industry. The problem is formulated as a general semi-infinite program (GSIP) and solved using an interior-point method developed by Stein [O. Stein, Bi-level Strategies in Semi-infinite Programming, Kluwer Academic Publishers, Boston, 2003]. It is shown, that the convexity assumption needed for the convergence of the algorithm can be satisfied by appropriate modelling. Clustering techniques are used to reduce the number of container constraints, which is necessary to make the subproblems practically tractable. An iterative process consisting of GSIP optimization and adaptive refinement steps is then employed to obtain an optimal solution which is also feasible for the original problem. Some numerical results based on real-world data are also presented.     

A dynamic programming extension to the steady state refinery-LP

Standard LPs, widely used for planning in the processing industry, are powerful tools for making economic decisions, but do not cover time relationships inside the planning timeframe. To include sequence-dependent issues within the optimization process, an algorithmic extension to the LP was developed. DPX (Dynamic Programming Extension) is used to solve a whole class of problems that are dynamic in time.     

A dual strategy for solving the linear programming relaxation of a driver scheduling system

A Mathematical Programming model of a driver scheduling system is described. This consists of set covering and partitioning constraints, possibly user-supplied side constraints, and two pre-emptively ordered objectives. The previous solution strategy addressed the two objectives using separate Primal Simplex optimisations; a new strategy uses a single weighted objective function and a Dual Simplex algorithm initiated by a specially developed heuristic. Computational results are reported.     

警力分配问题中的模糊优选动态规划法

根据市区公安分局基层警力的分配问题的特点,将模糊优选理论与动态规划法相结合建立了多维多目标模糊优选动态规划模型,用此模型对警力进行合理分配,以得到最优警力分配方案,为公安决策者在警力分配问题中提供科学的理论依据.     

连续型动态规划在投资决策中的应用

首先简要介绍了动态规划方法及在投资决策中的应用.然后给出一个具体示例进行分析和计算.     

Generalized Convex Disjunctive Programming: Nonlinear Convex Hull Relaxation

Generalized Disjunctive Programming (GDP) has been introduced recently as an alternative to mixed-integer programming for representing discrete/continuous optimization problems. The basic idea of GDP consists of representing these problems in terms of sets of disjunctions in the continuous space, and logic propositions in terms of Boolean variables. In this paper we consider GDP problems involving convex nonlinear inequalities in the disjunctions. Based on the work by Stubbs and Mehrotra [21] and Ceria and Soares [6], we propose a convex nonlinear relaxation of the nonlinear convex GDP problem that relies on the convex hull of each of the disjunctions that is obtained by variable disaggregation and reformulation of the inequalities. The proposed nonlinear relaxation is used to formulate the GDP problem as a Mixed-Integer Nonlinear Programming (MINLP) problem that is shown to be tighter than the conventional big-M formulation. A disjunctive branch and bound method is also presented, and numerical results are given for a set of test problems.     

Approximate dynamic programming with Bézier Curves/Surfaces for Top-percentile Traffic Routing

Multi-homing is used by Internet Service Providers (ISPs) to connect to the Internet via different network providers. This study develops a routing strategy under multi-homing in the case where network providers charge ISPs according to top-percentile pricing (i.e. based on the θth highest volume of traffic shipped). We call this problem the Top-percentile Traffic Routing Problem (TpTRP).Solution approaches based on Stochastic Dynamic Programming require discretization in state space, which introduces a large number of state variables. This is known as the curse of dimensionality in state space. To overcome this, in previous work we have suggested to use approximate dynamic programming (ADP) to construct value function approximations, which allow us to work in continuous state space. The resulting ADP model provides well performing routing policies for medium sized instances of the TpTRP. In this work we extend the ADP model, by using Bézier Curves/Surfaces to obtain continuous-time approximations of the time-dependent ADP parameters. This modification reduces the number of regression parameters to estimate, and thus accelerates the efficiency of parameter training in the solution of the ADP model, which makes realistically sized TpTRP instances tractable. We argue that our routing strategy is near optimal by giving bounds.     

价格控制问题的基本性质

价格控制问题是一类重要的二层规划问题。最近,文[3,4]讨论了价格控制问题的最优性条件和解集的性质。本文先用反例说明[3]中关于价格控制问题的可行解的充分必要条件的一个命题是不确切的,并对[4]中基于该命题证得的不可靠结论在适当条件下进行了证明。     

Resolving degeneracy in quadratic programming

A technique for the resolution of degeneracy in an Active Set Method for Quadratic Programming is described. The approach generalises Fletcher's method [2] which applies to the LP case. The method is described in terms of an LCP tableau, which is seen to provide useful insights. It is shown that the degeneracy procedure only needs to operate when the degenerate constraints are linearly dependent on those in the active set. No significant overheads are incurred by the degeneracy procedure. It is readily implemented in a null space format, and no complications in the matrix algebra are introduced.The guarantees of termination provided by [2], extending in particular to the case where round-off error is present, are preserved in the QP case. It is argued that the technique gives stronger guarantees than are available with other popular methods such as Wolfe's method [11] or the method of Goldfarb and Idnani [7].Presented at the 14th International Symposium on Mathematical Programming, Amsterdam, August 5–9, 1991.     

Clutter nonidealness

Several key results for set packing problems do not seem to be easily or even possibly transferable to set covering problems, although the symmetry between them. The goal of this paper is to introduce a nonidealness index by transferring the ideas used for the imperfection index defined by Gerke and McDiarmid [Graph imperfection, J. Combin. Theory Ser. B 83 (2001) 58-78]. We found a relationship between the two indices and the strength of facets defined in [M. Goemans, Worst-case comparison of valid inequalities for the TSP, mathematical programming, in: Fifth Integer Programming and Combinatorial Optimization Conference, Lecture Notes in Computer Science, vol. 1084, Vancouver, Canada, 1996, pp. 415-429; M. Goemans, L.A. Hall, The strongest facets of the acyclic subgraph polytope are unknown, in: Integer Programming and Combinatorial Optimization, Lecture Notes in Computer Science, vol. 1084, Springer, Berlin, 1996, pp. 415-429]. We prove that a clutter is as nonideal as its blocker and find some other properties that could be transferred from the imperfection index to the nonidealness index. Finally, we analyze the behavior of the nonidealness index under some clutter operations.     

On the set covering polyhedron of circulant matrices

A well known family of minimally nonideal matrices is the family of the incidence matrices of chordless odd cycles. A natural generalization of these matrices is given by the family of circulant matrices. Ideal and minimally nonideal circulant matrices have been completely identified by Cornuéjols and Novick [G. Cornuéjols, B. Novick, Ideal 0 - 1 matrices, Journal of Combinatorial Theory B 60 (1994) 145–157]. In this work we classify circulant matrices and their blockers in terms of the inequalities involved in their set covering polyhedra. We exploit the results due to Cornuéjols and Novick in the above-cited reference for describing the set covering polyhedron of blockers of circulant matrices. Finally, we point out that the results found on circulant matrices and their blockers present a remarkable analogy with a similar analysis of webs and antiwebs due to Pêcher and Wagler [A. Pêcher, A. Wagler, A construction for non-rank facets of stable set polytopes of webs, European Journal of Combinatorics 27 (2006) 1172–1185; A. Pêcher, A. Wagler, Almost all webs are not rank-perfect, Mathematical Programming Series B 105 (2006) 311–328] and Wagler [A. Wagler, Relaxing perfectness: Which graphs are ‘Almost’ perfect?, in: M. Groetschel (Ed.), The Sharpest Cut, Impact of Manfred Padberg and his work, in: SIAM/MPS Series on Optimization, vol. 4, Philadelphia, 2004; A. Wagler, Antiwebs are rank-perfect, 4OR 2 (2004) 149–152].