Routing Optimization with Generalized Consistency Requirements

The consistent vehicle routing problem (ConVRP) aims to design synchronized routes on multiple days to serve a group of customers while minimizing the total travel cost. It stipulates that customers should be visited at roughly the same time (time consistency) by several familiar drivers (driver consistency). This paper generalizes the ConVRP for any level of driver consistency and additionally addresses route consistency, which means that each driver can traverse at most a certain proportion of different arcs of routes on planning days, which guarantees route familiarity. To solve this problem, we develop two set partitioning-based formulations, one based on routes and the other based on schedules. We investigate valid lower bounds on the linear relaxations of both of the formulations that are used to derive a subset of columns (routes and schedules); within the subset are columns of an optimal solution for each formulation. We then solve the reduced problem of either one of the formulations to achieve an optimal solution. Numerical results show that our exact method can effectively solve most of the medium-sized ConVRP instances in the literature and can also solve some newly generated instances involving up to 50 customers. Our exact solutions explore some managerial findings with respect to the adoption of consistency measures in practice. First, maintaining reasonably high levels of consistency requirements does not necessarily always lead to a substantial increase in cost. Second, a high level of time consistency can potentially be guaranteed by adopting a high level of driver consistency. Third, maintaining high levels of time consistency and driver consistency may lead to lower levels of route consistency.

On new methods to construct lower bounds in simplicial branch and bound based on interval arithmetic

Branch and Bound (B&B) algorithms in Global Optimization are used to perform an exhaustive search over the feasible area. One choice is to use simplicial partition sets. Obtaining sharp and cheap bounds of the objective function over a simplex is very important in the construction of efficient Global Optimization B&B algorithms. Although enclosing a simplex in a box implies an overestimation, boxes are more natural when dealing with individual coordinate bounds, and bounding ranges with Interval Arithmetic (IA) is computationally cheap. This paper introduces several linear relaxations using gradient information and Affine Arithmetic and experimentally studies their efficiency compared to traditional lower bounds obtained by natural and centered IA forms and their adaption to simplices. A Global Optimization B&B algorithm with monotonicity test over a simplex is used to compare their efficiency over a set of low dimensional test problems with instances that either have a box constrained search region or where the feasible set is a simplex. Numerical results show that it is possible to obtain tight lower bounds over simplicial subsets.

On refinement strategies for solving $${\textsc {MINLP}\mathrm{s}}$$ by piecewise linear relaxations: a generalized red refinement

We investigate the generalized red refinement for n-dimensional simplices that dates back to Freudenthal (Ann Math 43(3):580–582, 1942) in a mixed-integer nonlinear program ($${\textsc {MINLP}}$$ MINLP ) context. We show that the red refinement meets sufficient convergence conditions for a known $${\textsc {MINLP}}$$ MINLP solution framework that is essentially based on solving piecewise linear relaxations. In addition, we prove that applying this refinement procedure results in piecewise linear relaxations that can be modeled by the well-known incremental method established by Markowitz and Manne (Econometrica 25(1):84–110, 1957). Finally, numerical results from the field of alternating current optimal power flow demonstrate the applicability of the red refinement in such $${\textsc {MIP}}$$ MIP -based $${\textsc {MINLP}}$$ MINLP solution frameworks.

Complexity of linear relaxations in integer programming

For a setXof integer points in a polyhedron, the smallest number of facets of any polyhedron whose set of integer points coincides with Xis called the relaxation complexity $${{\,\mathrm{rc}\,}}(X)$$rc(X). This parameter, introduced by Kaibel & Weltge (2015), captures the complexity of linear descriptions of Xwithout using auxiliary variables. Using tools from combinatorics, geometry of numbers, and quantifier elimination, we make progress on several open questions regarding$${{\,\mathrm{rc}\,}}(X)$$rc(X)and its variant$${{\,\mathrm{rc}\,}}_\mathbb {Q}(X)$$rcQ(X), restricting the descriptions of Xto rational polyhedra. As our main results we show that$${{\,\mathrm{rc}\,}}(X) = {{\,\mathrm{rc}\,}}_\mathbb {Q}(X)$$rc(X)=rcQ(X)when: (a)Xis at most four-dimensional, (b)Xrepresents every residue class in$$(\mathbb {Z}/2\mathbb {Z})^d$$(Z/2Z)d, (c) the convex hull of Xcontains an interior integer point, or (d) the lattice-width of Xis above a certain threshold. Additionally,$${{\,\mathrm{rc}\,}}(X)$$rc(X)can be algorithmically computed when Xis at most three-dimensional, orXsatisfies one of the conditions (b), (c), or (d) above. Moreover, we obtain an improved lower bound on$${{\,\mathrm{rc}\,}}(X)$$rc(X)in terms of the dimension of X.

On the Consistent Path Problem

This paper studies a novel decomposition scheme, utilizing decision diagrams for modeling elements of a problem where typical linear relaxations fail to provide sufficiently tight bounds. Given a collection of decision diagrams, each representing a portion of the problem, together with linear inequalities modeling other portions of the problem, how can one efficiently optimize over such a representation? In this paper, we model the problem as a consistent path problem, where a path in each diagram has to be identified, all of which agree on the value assignments to variables. We establish complexity results and propose a branch-and-cut framework for solving the decomposition. Through application to binary cubic optimization and a variant of the market split problem, we show that the decomposition approach provides significant improvement gains over standard linear models.

The Vehicle Routing Problem with Partial Outsourcing

This paper introduces the vehicle routing problem with partial outsourcing (VRPPO) in which a customer can be served by a single private vehicle, by a common carrier, or by both a single private vehicle and a common carrier. As such, it is a variant of the vehicle routing problem with private fleet and common carrier (VRPPC). The objective of the VRPPO is to minimize fixed and variable costs of the private fleet plus the outsourcing cost. We propose two different path-based formulations for the VRPPO and solve these with a branch-and-price-and-cut solution method. For each path-based formulation, two different pricing procedures are designed and used when solving the linear relaxations by column generation. To assess the quality of the solution methods and gain insight in potential cost improvements compared with the VRPPC, we perform tests on two instance sets with up to 100 customers from the literature.

Using multiflow formulations to solve the Steiner tree problem in graphs

We present three dfferent mixed integer linear models with a polynomial number of variables and constraints for the Steiner tree problem in graphs. The linear relaxations of these models are compared to show that a good (strong) linear relaxation can be a good approximation for the problem. We present computational results for the the STP OR-Library (J.E. Beasley) instances of type b, c, d and e.

Visual e-commerce values filtering framework with spatial database metric

Our customer preference model is based on aggregation of partly linear relaxations of value filters often used in e-commerce applications. Relaxation is motivated by the Analytic Hierarchy Processing method and combining fuzzy information in web accessible databases. In low dimensions our method is well suited also for data visualization. The process of translating models (user behavior) to programs (learned recommendation) is formalized by Challenge-Response Framework ChRF. ChRF resembles remote process call and reduction in combinatorial search problems. In our case, the model is automatically translated to a program using spatial database features. This enables us to define new metrics with visual motivation. We extend the conference paper with inductive ChRF, new representation of user and an additional method and metric. We provide experiments with synthetic data (items) and users.