The search result changed since you submitted your search request. Documents might be displayed in a different sort order.
  • search hit 1 of 2
Back to Result List

Solving Binary-Constrained Mixed Complementarity Problems Using Continuous Reformulations

  • Mixed complementarity problems are of great importance in practice since they appear in various fields of applications like energy markets, optimal stopping, or traffic equilibrium problems. However, they are also very challenging due to their inherent, nonconvex structure. In addition, recent applications require the incorporation of integrality constraints. Since complementarity problems often model some kind of equilibrium, these recent applications ask for equilibrium points that additionally satisfy certain integer conditions. Obviously, this makes the problem even harder to solve. The solution approach used most frequently in the literature is to recast the complementarity conditions as disjunctive constraints using additional binary variables and big-M constraints. However, both latter aspects create issues regarding the tractability and correctness of the reformulation. In this paper, we follow the opposite route and restate the integrality conditions as complementarity constraints, leading to purely continuous reformulations that can be tackled by local solvers. We study these reformulations theoretically and provide a numerical study that shows that continuous reformulations are useful in practice both in terms of solution times and solution quality.

Download full text files

Export metadata

Metadaten
Author:Steven A. Gabriel, Marina Leal, Martin Schmidt
Parent Title (German):Computers & Operations Research
Document Type:Article
Language:English
Date of Publication (online):2020/02/16
Date of first Publication:2020/02/16
Release Date:2020/02/16
Tag:Binary-constrained mixed complementarity problems; Continuous reformulations; Mixed-integer optimization; Spatial price equilibrium problems
Page Number:24
Institutes:Universität Trier
Subprojects:A05
B08
Licence (German):License LogoCreative Commons - CC BY - Namensnennung 4.0 International