The inverse 1-median location problem on uncertain tree networks with tail value at risk criterion
Introduction
One of the important aspects of the inverse optimization which has recently been studied by many researchers is the inverse location problem, in which the task is to change the parameters of the original problem at the minimum total cost with respect to modification bounds such that the given facility location under the new values of the parameters becomes optimal. One of the well-known models is the inverse 1-median location problem (I1-MLP) which is stated on a tree network as follows:
Consider as a tree network with vertex set and edge set E. Let d(vi, vj) be the distance between two vertices vi and vj in T. Moreover, let wi be a nonnegative weight associated to vi ∈ V, and each edge e ∈ E has a positive length. We call a vertex v* as a 1-median of the tree network T if and only if for all vj ∈ V the inequalityis satisfied.
Assume that vi, are the immediate neighbors of an arbitrary vertex v*. Denote by the subtrees rooted at the vertices respectively. DefineGoldman [10] proved that a vertex v* ∈ V is a 1-median of the tree network T, if and only if
In an I1-MLP on a tree network, the task is to modify the vertex weights at the minimum total cost such that a given vertex becomes a 1-median of the tree network with respect to new vertex weights. Let the vertex weights obey the upper and lower bounds and wi, respectively. Suppose that v* is not a 1-median with respect to the original vertex weights. In an I1-MLP model, the task is to modify wi to for all vi ∈ V such that
- 1)
The vertex v* ∈ V becomes a 1-median of the tree with respect to .
- 2)
The following bound constraints are satisfied:
- 3)
The objective cost function becomes minimum under a specific norm.
Assume that we incur the nonnegative costs and if the vertex weight of vi is increased and decreased by one unit, respectively. Define if wi is increased and if wi is decreased. The total modification costs can be measured by the weighted l1 norm (Wl1N), weighted l∞ norm (Wl∞N) and weighted bottleneck-Hamming distance (WBHD) which are stated as follows, respectively:where if and 0 otherwise.
The literature review shows that the inverse location problems have recently been studied by many researchers. For a survey on inverse 1-median location problems, see e.g., [1], [4], [5], [9], [11], [12], [13], [20], [24].
In the real life, we are usually faced with various types of uncertainty. For example, some parameters of the location problems like, the vertex weights, the travel times between vertices, the establishing costs of facilities and the network modification costs may not be certainly known. The uncertainty theory introduced by Liu [16] is a suitable tool to deal with such parameters. Gao [7] modeled the single facility location problems with uncertain demands. Wen et al. [29] investigated the capacitated facility location-allocation problem with uncertain demands. Nguyen and Chi [21] studied I1-MLP on a tree with uncertain costs and showed that the inverse distribution function of the minimum cost can be obtained at O(n2log n) time. Zhang et al. [32] investigated the sustainable multi-depot emergency facilities location-routing problem with uncertain information. For a survey on uncertain location problems, we refer the interested reader to [8], [14], [15], [19], [23], [25], [26], [31]. Forthermore, see [6] as a related work.
Note that the uncertainty leads to risk. Liu [18] introduced the risk concept in the uncertain environment. Measuring the risk is one of the important steps in the decision making process. The risk metrics contain techniques and data sets used to calculate the risk value of the problem under investigation. Tail value at risk (TVaR) metric [22] is one of the measures of the risk that is widely acceptable among industry segments and market participants. For a survey on the risk management in the location problems with random and fuzzy variables, see e.g., [2], [3], [27], [28], [30].
In this paper, we concentrate on the I1-MLP model in a tree network with uncertain vertex weights and weight modification costs under Wl1N. The article is organized as follows: In the next section, we first present some basic definitions and theorems in the uncertainty theory. Then, the TVaR metric and its property in an uncertain environment are introduced. In Section 3, we derive a necessary and sufficient condition for the α-1-median on an uncertain tree and apply this condition to propose an equivalent model with deterministic constraints for the uncertain inverse 1-median location problem (UI1-MLP). Then, we consider the vertex weights as linear uncertain variables and develop a new model for the problem. Since the costs are considered as uncertain variables, we solve the obtained model with TVaR objective. Finally, we suppose that the uncertain costs have linear or zigzag distributions and then we present a combinatorial algorithm with time complexity of O(nlog n) for the problem under Wl1N. The conclusion of the paper is given in Section 4.
Section snippets
Preliminaries
In this section, we present some definitions and theorems of the uncertainty theory and the TVaR metric.
I1-MLP on uncertain tree networks
In this section, I1-MLP on a tree network with uncertain vertex weights and uncertain costs is investigated and a novel combinatorial solution algorithm is provided.
Let θi, vi ∈ V, be the independent uncertain vertex weights on the tree T related to the associated parameters wi, vi ∈ V. In fact, the original weights will be denoted by θi(wi) and the new weights will be denoted by with respect to the modified parameters . Moreover, let and be the independent uncertain
Conclusion
Inverse 1-median location problem in uncertain environments has been investigated by Nguyen and Chi [21] and Soltanpour et al. [25]. Nguyen and Chi [21] studied the inverse 1-median location problem on tree networks with uncertain costs and showed that the inverse distribution function of the minimum cost can be obtained in O(n2log n) time. Soltanpour et al. [25] investigated the inverse 1-median location problem on tree networks with intuitionistic fuzzy costs and obtained its value at risk
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (32)
- et al.
Location choice and risk attitude of a decision maker
Omega
(2017) - et al.
Inverse median problems
Discret. Optim.
(2004) - et al.
The inverse 1-median problem on a cycle
Discret. Optim.
(2008) - et al.
Uncertain gaol programming models for bicriteria solid transportation problem
Appl. Soft Comput.
(2017) Uncertain models for single facility location problems on networks
Appl. Math. Model.
(2012)- et al.
A chance constrained programming approach for uncertain p-hub center location problem
Comput. Ind. Eng.
(2016) The inverse 1-median problem on a tree and on a path
Electron. Notes Discret. Math.
(2010)2-balanced flows and the inverse 1-median problem in the Chebyshev space
Discret. Optim.
(2012)Uncertain risk analysis and uncertain reliability analysis
J. Uncertain Syst.
(2010)- et al.
Risk management in uncapacitated facility location models with random demands
Comput. Oper. Res.
(2009)
Optimizing fuzzy p-hub center problem with generalized value at risk criterion
Appl. Math. Model.
Covering location problem of emergency service facilities in an uncertain environment
Appl. Math. Model.
Sustainable multi-depot emergency facilities location-routing problem with uncertain information
Appl. Math. Comput.
Inverse p-median problems with variable edge lengths
Math. Methods Oper. Res.
Minimum risk facility location-allocation problem with type-2 fuzzy variables
Sci. World J.
Optimal center location in simple networks
Transp. Sci.
Cited by (8)
Mean–variance value at risk criterion for solving a p-median location problem on networks with type-2 intuitionistic fuzzy weights
2024, Journal of Computational and Applied MathematicsThe minmax regret inverse maximum weight problem
2021, Applied Mathematics and ComputationCitation Excerpt :Nguyen and Chi [26] was the first who proposed a model for inverse 1-median problem on trees in term of uncertainty theory. In addition, Soltanpour [29,30] investigated the same problem under the fuzziness of cost coefficients. If the exact distribution of an uncertain variable is not known so far, we apply the concept of robustness to measure the resulting cost function.
Computational aspects of the inverse single facility location problem on trees under l<inf>k</inf>-norm
2020, Theoretical Computer ScienceInverse Vertex/Absolute Quickest 1-Center Location Problem on a Tree Under Weighted l<inf>1</inf> Norm
2024, Journal of Optimization Theory and ApplicationsDesigning a Railway Network in Cesme, Izmir with Bi-objective Ring Star Problem
2022, Lecture Notes in Mechanical Engineering