Dual Optimal Solutions Research Articles

A parametric programming approach to bilevel optimisation with lower-level variables in the upper level

This paper examines linearly constrained bilevel programming problems in which the upper-level objective function depends on both the lower-level primal and dual optimal solutions. We parametrize the lower-level solutions and thereby the upper-level objective function by the upper-level variables and argue that it may be non-convex and even discontinuous. However, when the upper-level objective is affine in the lower-level primal optimal solution, the parametric function is piece-wise linear. We show how this property facilitates the application of parametric programming and demonstrate how the approach allows for decomposition of a separable lower-level problem. When the upper-level objective is bilinear in the lower-level primal and dual optimal solutions, we also provide an exact linearisation method that reduces the bilevel problem to a single-level mixed-integer linear programme (MILP). We assess the performance of the parametric programming approach on two case studies of strategic investment in electricity markets and benchmark against state-of-the-art MILP and non-linear solution methods for bilevel optimisation problems. Preliminary results indicate substantial computational advantages over several standard solvers, especially when the lower-level problem separates into a large number of subproblems. Furthermore, we show that the parametric programming approach succeeds in solving problems to global optimality for which standard methods can fail.

Disruption Risk Mitigation in Supply Chains: The Risk Exposure Index Revisited

In recent years, supply chains are more prone to disruptions. The impact on performance depends on the system's ability to discover and then recover after the disruption has occurred. In this paper, we proposed a new method to integrate probabilistic assessment of disruption risks into the Risk Exposure Index (REI) approach proposed previously by Simchi-Levi et al. and measure supply chain resiliency by analyzing the worst-case CVaR (WCVaR) of total lost sales under disruptions. We show that the optimal strategic inventory positioning strategy in this model can be fully characterized by a conic program. The optimal primal and dual solutions to the conic program can be used to shed light on comparative statics in the supply chain risk mitigation problem. This information can help supply chain risk managers focus their mitigation efforts on critical suppliers and/or installations that will have greater impact on the performance of the supply chain when disrupted.

Cost allocation for rural electrification using game theory: A case of distributed generation in rural India

This paper deals with an economic aspect of distributed generation systems which rural energy cooperatives may mainly support, since they can effectively coordinate the different stakeholders within an energy access initiative. Members of a rural energy cooperative would share the costs of deploying a distributed generation system, so that each of them can benefit from it. To minimise the overall investment cost, such a cooperative must consider both technical sizing and cost allocation among its members. For the latter, an important issue to take into account is the so-called willingness-to-pay. For obtaining the design of the distributed generation system we propose to use facility location covering (FALCO) models. The solution of the FALCO model provides both the structure of the distributed generation system and the investment cost to be covered by the members of the cooperative. We take advantage of the structure of the primal and dual optimal solutions of the FALCO model to endogenously determine the minimum and maximum willingness-to-pay of the cooperative members and then to apply rationing techniques to obtain the final allocation of costs to each member. Thus, our approach optimises the generator locations and minimises the total system cost, while assigning to each member at least her minimum willingness-to-pay. From our point of view, this method may help the beneficiaries to accept the investment cost allocation. Finally, a case study based on real data from a rural village in Maharashtra, where a group of farmers wanting to buy and share an energy system to pump water for irrigating their fields, is used to illustrate the working of the proposed cost allocation methodology.

A Primal-dual Learning Algorithm for Personalized Dynamic Pricing with an Inventory Constraint

We consider the problem of a firm seeking to use personalized pricing to sell an exogenously given stock of a product over a finite selling horizon to different consumer types. We assume that the type of an arriving consumer can be observed but the demand function associated with each type is initially unknown. The firm sets personalized prices dynamically for each type and attempts to maximize the revenue over the season. We provide a learning algorithm that is near-optimal when the demand and capacity scale in proportion. The algorithm utilizes the primal-dual formulation of the problem and learns the dual optimal solution explicitly. It allows the algorithm to overcome the curse of dimensionality (the rate of regret is independent of the number of types) and sheds light on novel algorithmic designs for learning problems with resource constraints.

An integer optimality condition for column generation on zero–one linear programs

Column generation is a linear programming method that, when combined with appropriate integer programming techniques, has been successfully used for solving huge integer programs. The method alternates between a restricted master problem and a column generation subproblem. The latter step is founded on dual information from the former one; often an optimal dual solution to the linear programming relaxation of the restricted master problem is used.We consider a zero–one linear programming problem that is approached by column generation and present a generic sufficient optimality condition for the restricted master problem to contain the columns required to find an integer optimal solution to the complete problem. The condition is based on dual information, but not necessarily on an optimal dual solution. It is however most natural to apply the condition in a situation when an optimal or near-optimal dual solution is at hand.We relate our result to a few special cases from the literature, and make some suggestions regarding possible exploitation of the optimality condition in the construction of column generation methods for integer programs.

Delay-Optimal Proactive Service Framework for Block-Stream as a Service

Block-stream as a service (BaaS) has been considered as a promising service provisioning model for lightweight devices, in which the server streams code blocks partitioned from an APP to devices in an on-demand way. However, serving with code blocks makes the quality-of-services of BaaS more prone to the dynamics in users’ requests and wireless links. In this letter, we propose a proactive service framework to minimize the response delay while bounding the time-average energy consumption, by proactively streaming blocks before the requests. The framework is composed of three components, i.e., model estimation, dual learning, and block serving. The first two components achieve the statistics of the request arrivals and the optimal dual solution of the original stochastic problem, respectively. The last component determines whether to proactively serve blocks based only on the instantaneous and estimated system states.

The complexity of primal-dual fixed point methods for ridge regression

We study the ridge regression (L2 regularized least squares) problem and its dual, which is also a ridge regression problem. We observe that the optimality conditions describing the primal and dual optimal solutions can be formulated in several different but equivalent ways. The optimality conditions we identify form a linear system involving a structured matrix depending on a single relaxation parameter which we introduce for regularization purposes. This leads to the idea of studying and comparing, in theory and practice, the performance of the fixed point method applied to these reformulations. We compute the optimal relaxation parameters and uncover interesting connections between the complexity bounds of the variants of the fixed point scheme we consider. These connections follow from a close link between the spectral properties of the associated matrices. For instance, some reformulations involve purely imaginary eigenvalues; some involve real eigenvalues and others have all eigenvalues on the complex circle. We show that the deterministic Quartz method—which is a special case of the randomized dual coordinate ascent method with arbitrary sampling recently developed by Qu, Richtárik and Zhang—can be cast in our framework, and achieves the best rate in theory and in numerical experiments among the fixed point methods we study. Remarkably, the method achieves an accelerated convergence rate. Numerical experiments indicate that our main algorithm is competitive with the conjugate gradient method.

The fixed charge transportation problem: a strong formulation based on Lagrangian decomposition and column generation

A new and strong convexified formulation of the fixed charge transportation problem is provided. This formulation is obtained by integrating the concepts of Lagrangian decomposition and column generation. The decomposition is made by splitting the shipping variables into supply and demand side copies, while the columns correspond to extreme flow patterns for single sources or sinks. It is shown that the combination of Lagrangian decomposition and column generation provides a formulation whose strength dominates those of three other convexified formulations of the problem. Numerical results illustrate that our integrated approach has the ability to provide strong lower bounds. The Lagrangian decomposition yields a dual problem with an unbounded set of optimal solutions. We propose a regularized column generation scheme which prioritizes an optimal dual solution with a small $$l_1$$ -norm. We further demonstrate numerically that information gained from the strong formulation can also be used for constructing a small-sized core problem which yields high-quality upper bounds.

A primal-dual homotopy algorithm for $$\ell _{1}$$-minimization with $$\ell _{\infty }$$-constraints

In this paper we propose a primal-dual homotopy method for $$\ell _1$$ -minimization problems with infinity norm constraints in the context of sparse reconstruction. The natural homotopy parameter is the value of the bound for the constraints and we show that there exists a piecewise linear solution path with finitely many break points for the primal problem and a respective piecewise constant path for the dual problem. We show that by solving a small linear program, one can jump to the next primal break point and then, solving another small linear program, a new optimal dual solution is calculated which enables the next such jump in the subsequent iteration. Using a theorem of the alternative, we show that the method never gets stuck and indeed calculates the whole path in a finite number of steps. Numerical experiments demonstrate the effectiveness of our algorithm. In many cases, our method significantly outperforms commercial LP solvers; this is possible since our approach employs a sequence of considerably simpler auxiliary linear programs that can be solved efficiently with specialized active-set strategies.

An adaptive line search scheme for approximated nuclear norm based matrix regression

Abstract Face recognition with partial occlusion is common in the real world and has become a hot topic in the pattern recognition research. Recently, nuclear norm based matrix regression model (NMR) has been shown a great potential in dealing with the structural noise. And yet, it uses Alternating Direction Method of Multipliers (ADMM) to achieve the optimal regression coefficients, which needs to bring into an auxiliary variable and only exploits the convexity of NMR. Comparing with ADMM, the gradient based methods are simpler. To take advantage of these methods, this paper considers the Approximated NMR model with Elastic-net regularization (ANMRE). By virtue of the soft-thresholding (shrinkage) operator, singular value shrinkage operator and strong convexity of ANMRE, the dual problem of ANMRE (DANMRE) is derived and a crucial result is obtained: the primal optimal solution of ANMRE can be converted as the matrix function associated with the dual optimal solution. Leveraging the differentiability of DANMRE, an adaptive line search scheme is developed to solve it. This approach makes full use of the advantages of the accelerated gradient technique and adaptive parameters updating strategy. Meanwhile, a convergence rate of O(1/N2) can be guaranteed under mild conditions. Experimental results on five standard databases show the superiority of the proposed algorithm over some existing methods.

Ensembles of Lasso Screening Rules.

In order to solve large-scale lasso problems, screening algorithms have been developed that discard features with zero coefficients based on a computationally efficient screening rule. Most existing screening rules were developed from a spherical constraint and half-space constraints on a dual optimal solution. However, existing rules admit at most two half-space constraints due to the computational cost incurred by the half-spaces, even though additional constraints may be useful to discard more features. In this paper, we present AdaScreen, an adaptive lasso screening rule ensemble, which allows to combine any one sphere with multiple half-space constraints on a dual optimal solution. Thanks to geometrical considerations that lead to a simple closed form solution for AdaScreen, we can incorporate multiple half-space constraints at small computational cost. In our experiments, we show that AdaScreen with multiple half-space constraints simultaneously improves screening performance and speeds up lasso solvers.

Convergence of a proximal algorithm for solving the dual of a generalized fractional program

We propose to use the proximal point algorithm to regularize a “dual” problem of generalized fractional programs (GFP). The proposed technique leads to a new dual algorithm that generates a sequence which converges from below to the minimal value of the considered problem. At each step, the proposed algorithm solves approximately an auxiliary problem with a unique dual solution whose every cluster point gives a solution to the dual problem. In the exact minimization case, the sequence of dual solutions converges to an optimal dual solution. For a class of functions, including the linear case, the convergence of the dual values is at least linear.

Prize‐collecting set multicovering with submodular pricing

AbstractWe consider a ground set E and a submodular function acting on it. We first propose a “set multicovering” problem in which the value (price) of any is . We show that the linear program (LP) of this problem can be directly solved by applying a submodular function minimization (SFM) algorithm on the dual LP. However, the main results of this study concern prize‐collecting multicovering with submodular pricing, that is, a more general and more difficult “multicovering” problem in which elements can be left uncovered by paying a penalty. We formulate it as a large‐scale LP (with variables representing all subsets of E) that could be tackled by column generation (CG; for a CG algorithm for “set‐covering” problems with submodular pricing). However, we do not solve this large‐scale LP by CG, but we solve it in polynomial time by exploiting its integrality properties. More exactly, after appropriate restructuring, the dual LP can be transformed into an LP that has been thoroughly studied (as a primal) in the SFM theory. Solving this LP reduces to optimizing a strong map of submodular functions. For this, we use the Fleischer–Iwata framework that optimizes all these functions within the same asymptotic running time as solving a single SFM, that is, in , where and γ is the complexity of one submodular evaluation. Besides solving the problem, the proposed approach can be useful to (a) simultaneously find the best solution of up to functions under strong map relations in time, (b) perform sensitivity analysis in very short time (even at no extra cost), and (c) reveal combinatorial insight into the primal–dual optimal solutions. We present several potential applications throughout the paper, from production planning to combinatorial auctions.

New Algorithms for Maximum Weight Matching and a Decomposition Theorem

We revisit the classical maximum weight matching problem in general graphs with nonnegative integral edge weights. We present an algorithm that operates by decomposing the problem into W unweighted versions of the problem, where W is the largest edge weight. Our algorithm has running time as good as the current fastest algorithms for the maximum weight matching problem when W is small. One of the highlights of our algorithm is that it also produces an integral optimal dual solution; thus our algorithm also returns an integral certificate corresponding to the maximum weight matching that was computed. Our algorithm yields a new proof to the total dual integrality of Edmonds’ matching polytope and it also gives rise to a decomposition theorem for the maximum weight of a matching in terms of the maximum size of a matching in certain subgraphs. We also consider the maximum weight capacitated b-matching problem in bipartite graphs with nonnegative integral edge weights and show that it can also be decomposed into W unweighted versions of the problem, where W is the largest edge weight. Our second algorithm is competitive with known algorithms when W is small.

A Strongly Polynomial Algorithm for Generalized Flow Maximization

A strongly polynomial algorithm is given for the generalized flow maximization problem. It uses a new variant of the scaling technique called continuous scaling. The main measure of progress is that within a strongly polynomial number of steps, an arc can be identified that must be tight in every dual optimal solution and thus can be contracted. As a consequence of the result, we also obtain a strongly polynomial algorithm for the linear feasibility problem with at most two nonzero entries per column in the constraint matrix.

Stabilized column generation for the temporal knapsack problem using dual-optimal inequalities

We present two new methods to stabilize column-generation algorithms for the temporal knapsack problem (TKP). Caprara et al. (INFORMS J Comp 25(3):560–571, 2013] were the first to suggest the use of branch-and-price algorithms for Dantzig–Wolfe reformulations of the TKP. Herein, the respective pricing problems are smaller-sized TKP that can be solved with a general-purpose MIP solver or by dynamic programming. Our stabilization methods are tailored to the TKP as they use (deep) dual-optimal inequalities, that is, inequalities known to be fulfilled by all (at least some) optimal dual solutions to the linear relaxation. Extensive computational tests reveal that both new stabilization techniques are helpful. Several previously unsolved instances are now solved to proven optimality.

Distributed Robust Optimization in Networked System.

In this paper, we consider a distributed robust optimization (DRO) problem, where multiple agents in a networked system cooperatively minimize a global convex objective function with respect to a global variable under the global constraints. The objective function can be represented by a sum of local objective functions. The global constraints contain some uncertain parameters which are partially known, and can be characterized by some inequality constraints. After problem transformation, we adopt the Lagrangian primal-dual method to solve this problem. We prove that the primal and dual optimal solutions of the problem are restricted in some specific sets, and we give a method to construct these sets. Then, we propose a DRO algorithm to find the primal-dual optimal solutions of the Lagrangian function, which consists of a subgradient step, a projection step, and a diffusion step, and in the projection step of the algorithm, the optimized variables are projected onto the specific sets to guarantee the boundedness of the subgradients. Convergence analysis and numerical simulations verifying the performance of the proposed algorithm are then provided. Further, for nonconvex DRO problem, the corresponding approach and algorithm framework are also provided.

Pricing and clearing combinatorial markets with singleton and swap orders

In this article we consider combinatorial markets with valuations only for singletons and pairs of buy/sell-orders for swapping two items in equal quantity. We provide an algorithm that permits polynomial time market-clearing and -pricing. The results are presented in the context of our main application: the futures opening auction problem. Futures contracts are an important tool to mitigate market risk and counterparty credit risk. In futures markets these contracts can be traded with varying expiration dates and underlyings. A common hedging strategy is to roll positions forward into the next expiration date, however this strategy comes with significant operational risk. To address this risk, exchanges started to offer so-called futures contract combinations, which allow the traders for swapping two futures contracts with different expiration dates or for swapping two futures contracts with different underlyings. In theory, the price is in both cases the difference of the two involved futures contracts. However, in particular in the opening auctions price inefficiencies often occur due to suboptimal clearing, leading to potential arbitrage opportunities. We present a minimum cost flow formulation of the futures opening auction problem that guarantees consistent prices. The core ideas are to model orders as arcs in a network, to enforce the equilibrium conditions with the help of two hierarchical objectives, and to combine these objectives into a single weighted objective while preserving the price information of dual optimal solutions. The resulting optimization problem can be solved in polynomial time and computational tests establish an empirical performance suitable for production environments.

Duality Approaches to Economic Lot‐Sizing Games

Sharing common production, resources, and services to reduce cost are important for not for profit operations due to limited and mission‐oriented budget and effective cost allocation mechanisms are essential for encouraging effective collaborations. In this study, we illustrate how rigorous methodologies can be developed to derive effective cost allocations to facilitate sustainable collaborations in not for profit operations by modeling the cost allocation problem arising from an economic lot‐sizing (ELS) setting as a cooperative game. Specifically, we consider the economic lot‐sizing (ELS) game with general concave ordering cost. In this cooperative game, multiple retailers form a coalition by placing joint orders to a single supplier in order to reduce ordering cost. When both the inventory holding cost and backlogging cost are linear functions, it can be shown that the core of this game is non‐empty. The main contribution of this study is to show that a core allocation can be computed in polynomial time under the assumption that all retailers have the same cost parameters. Our approach is based on linear programming (LP) duality. More specifically, we study an integer programming formulation for the ELS problem and show that its LP relaxation admits zero integrality gap, which makes it possible to analyze the ELS game by using LP duality. We show that there exists an optimal dual solution that defines an allocation in the core. An interesting feature of our approach is that it is not necessarily true that every optimal dual solution defines a core allocation. This is in contrast to the duality approach for other known cooperative games in the literature.

Applications of gauge duality in robust principal component analysis and semidefinite programming

Gauge duality theory was originated by Freund [Math. Programming, 38(1):47-67, 1987] and was recently further investigated by Friedlander, Mac{\^e}do and Pong [SIAM J. Optm., 24(4):1999-2022, 2014]. When solving some matrix optimization problems via gauge dual, one is usually able to avoid full matrix decompositions such as singular value and/or eigenvalue decompositions. In such an approach, a gauge dual problem is solved in the first stage, and then an optimal solution to the primal problem can be recovered from the dual optimal solution obtained in the first stage. Recently, this theory has been applied to a class of \emph{semidefinite programming} (SDP) problems with promising numerical results [Friedlander and Mac{\^e}do, SIAM J. Sci. Comp., to appear, 2016]. In this paper, we establish some theoretical results on applying the gauge duality theory to robust \emph{principal component analysis} (PCA) and general SDP. For each problem, we present its gauge dual problem, characterize the optimality conditions for the primal-dual gauge pair, and validate a way to recover a primal optimal solution from a dual one. These results are extensions of [Friedlander and Mac{\^e}do, SIAM J. Sci. Comp., to appear, 2016] from nuclear norm regularization to robust PCA and from a special class of SDP which requires the coefficient matrix in the linear objective to be positive definite to SDP problems without this restriction. Our results provide further understanding in the potential advantages and disadvantages of the gauge duality theory.