Computational results data for the assoziated publication "Network Interdiction Problems in Urban Transportation: Theoretical Insights and Computational Characteristics"

This repository contains two Excel tables with computational results for our paper "Network Interdiction Problems in Urban Transportation: Theoretical Insights and Computational Characteristics". Each Excel table includes multiple worksheets, each representing different scenarios and models evaluated in our study.

Worksheets Overview

Each Excel table contains the following worksheets:
1. ML: Results for the "ML" big M values.
2. MH: Results for the "MH" big M values.
3. MF: Results for the "MF" big M values.
4. Path: Results for the path model.
5. FMInstances: Results from applying our models on the original Fontaine and Minner (2018) instances.

Columns Description

Each worksheet contains the following columns:

- Name of Instance: A complex string with the identifier of the used instance. The relevant part is "_RXXX_", where XXX is the random seed used to generate the instance.
- Number users: The number of users/commodities in the network.
- B: The budget (always set to infinity in our instances).
- GUROBI_RUNTIME: The time limit set for the computations.
- Modelkind: The type of model used. Possible values are:
  - INDICATOR: Compact model.
  - FMbenders: Benders model from Fontaine and Minner (2018).
  - ICM: Benders-like cuts.
  - PathModel: Path enumeration model.
- BigM computation: Time required to compute all the big M values used (not included in the time limit).
- runtime: Runtime of the selected model.
- BBnodes: Number of nodes in the Branch & Bound tree.
- gap: Gap reported by Gurobi after reaching the time limit.
- Cuts BLC: Number of Benders-like cuts included.
- Time BLC: Time required for separating Benders-like cuts.
- M improve BLC: Frequency of improvements to a big M when using the improved big M term in Benders-like cuts.
- Mcutoff_AVE: Average (non-zero) improvement of a big M when using the improved big M term in Benders-like cuts.
- Cuts FMBenders: Number of Benders cuts generated in the Fontaine and Minner (2018) model.
- Time FMBenders: Time required to generate the Benders cuts in the Fontaine and Minner (2018) model.
- Runtime path enum: Time required to enumerate all paths for the path-based model (not included in the time limit).
- Average Num Path: Average number of paths generated for a single commodity/user. Multiply this value by the number of users to obtain the absolute number of paths generated.

Note on FCP

All the results found for the instances already had integer flow solutions. Additionally, we conducted experiments where we explicitly forced the solutions to be integer for the Benders-like cuts model. We observed that the runtimes remained the same, with only some natural insignificant hardware-induced fluctuations. Therefore, we omit reporting these results again.

For further information or questions, please refer to our paper "Network Interdiction Problems in Urban Transportation: Theoretical Insights and Computational Characteristics" or contact the authors.