Constraint Solving Approaches to the Business-to-Business Meeting Scheduling Problem
Authors: Miquel Bofill, Jordi Coll, Marc Garcia, Jesús Giráldez-Cru, Gilles Pesant, Josep Suy, Mateu Villaret
JAIR 2022 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments conducted on real world instances, as well as on crafted ones, show that the Max SAT approach is the one with the best performance for this problem, exhibiting better solving times, sometimes even orders of magnitude smaller than CP and MIP. 7. Evaluation In this section we describe the results of the experimental investigation performed. |
| Researcher Affiliation | Academia | Miquel Bofill EMAIL IMAE department, University of Girona, Girona, Spain; Jordi Coll EMAIL Aix Marseille Univ, Universit e de Toulon, CNRS, LIS, Marseille, France; Marc Garcia EMAIL IMAE department, University of Girona, Girona, Spain; Jes us Gir aldez-Cru EMAIL Da SCI institute, DECSAI, University of Granada, Granada, Spain; Gilles Pesant EMAIL Polytechnique Montr eal, Montr eal, Canada; Josep Suy EMAIL Mateu Villaret EMAIL IMAE department, University of Girona, Girona, Spain |
| Pseudocode | Yes | Appendix. Mini Zinc Models In this appendix we provide Mini Zinc models for the CP approach. Namely, we fully provide and comment the Mini Zinc model using automaton A1 without meeting/no-meeting reification. Finally we also provide the changes needed in order to use this reification and automaton A0. Listing 1: Parameters... Listing 2: Variables... Listing 3: Feasibility Constraints... Listing 4: Optimization... Listing 5: Additional constraints... Listing 6: Search strategy for Chuffed... Listing 7: Automaton A0 with Meeting/No-meeting Reification... Listing 8: Minizinc code for Constraint (4)... Listing 9: Redundant Minizinc code for Constraint (5)... |
| Open Source Code | No | The paper provides Mini Zinc models in the appendix (Listings 1-9), which represent parts of the methodology. However, there is no unambiguous statement from the authors explicitly stating that they are releasing the full source code for the methodology described in the paper, nor a direct link to a code repository. |
| Open Datasets | Yes | 1. The instances are available at http://imae.udg.edu/recerca/lai. |
| Dataset Splits | No | The paper evaluates different solving approaches on a set of "original B2B instances" and "crafted instances" (modifications of the original ones). It describes how these crafted instances are generated (e.g., varying density of forbidden time slots α = 0.3% and α = 0.7%, fixed meetings β = {20, 40}%, precedences γ = {15, 25}%). However, there is no mention of traditional training/test/validation dataset splits used in machine learning or statistical modeling contexts. |
| Hardware Specification | Yes | The experiments have been run on a cluster of CPU nodes Intel Xeon E3-1220v2 at 3.10 GHz with 8GB of RAM. |
| Software Dependencies | Yes | These models have been implemented using Mini Zinc. Thanks to this we have been able to easily compare three solving methodologies: pure CP with Gecode2 (Schulte, Tack, & Lagerkvist, 2019) version 6.2.0, Lazy Clause Generation with Chuffed3 (Chu, 2011) commit 23b9fce, and integer programming with IBM ILOG Cplex4 (Cplex, 2009) version 12.9.0. The Max SAT solver used is UWr Max SAT6 (Piotr ow, 2019) version 1.0, which was one of the best Max SAT solvers of the unweighted track of the 2019 Max SAT Evaluation. |
| Experiment Setup | Yes | The timeout is 7200 seconds in all experiments. Regarding search strategies, we considered input order and first fail strategies for s variables... for Chuffed we obtained much better results with a search strategy that first tries to make the objective value as small as possible, and then continues with input order over variables s. Moreover, for Chuffed we also enable the free search option, which alternates between the given search strategy and activity-based search strategy. Listing 6: Search strategy for Chuffed. solve :: seq_search([int_search([cost], input_order, indomain_min),int_search(s, input_order, indomain_min)]) minimize cost; % search strategy and objective for Chuffed |