vc-optimizer

Detailed project report: Yannik_Kopyciok_VC_project_report.pdf

Description

This project contains code to run the integer linear programming (ILP) optimization described in the above mentioned report.

Packages

The following packages are required to run this project:

• gurobipy 10.0.1
• numpy 1.24.3
• scipy 1.10.1
• networkx 3.1
• matplotlib 3.7.1
• pandas 2.0.3

Files

The following files are included in this project:

• [fullILP.py]
• [fullILPUtils.py]
• [constants.py]
• [lightningMiniatureGen.py]
• [randGraphGenerator.py]

Walkthrough

ILP for a new set of LN miniature graphs

1. Register for a GurobiPy license (free for academia) and install the library as described on their website.
2. Install the required packages.
3. Run the ILP as follows:
   1. Open constants.py and change the GRAPH_SIZES_LB and GRAPH_SIZES_UB to the number of nodes you want to run the ILP on.
   2. Specify how many unique transactions should be constructed by adjusting the NUMBER_OF_TRXS variable (default: 5).
   3. Specify if and how often those transactions should be repeated by adjusting the REPETITIONS variable. Each number of repetition will create a set of transactions. For example, if you set REPETITIONS to 4 and NUMBER_OF_TRXS to 5, 4 sets of transactions will be created. The first contains only the 5 unique transactions, the second one contains the 5 unique transactions two times, and so on.
   4. Set GRAPH_TOPOLOGY to LN_MINIATURE.
   5. Open the file lightningMiniatureGen.py and run it (depending on lower and upper bound, this might take a while).
   6. Open the file fullILP.py and run it. When you created several repetitions, the ILP will always only run for one set of transactions. Which set it should run for can be adjusted with the variable NUMBER_OF_TRXS, which can be adjusted along the above example to 5, 10, 15, and 20.
   7. The results are saved in results.txt and can be visualized with resultGen.py. The default will plot the runtime of the prerequisites relative to the number of nodes.

ILP for the unique graph which is visualized in the project report with a set of 10 transactions (5 unique ones)

1. Change the variable GRAPH_TOPOLOGY in constants.py to UNIQ.
2. If you want the network to be visualized, change DRAWING to True. Remember to set it to False afterwards; otherwise, when running sets, every graph will be visualized.
3. Open fullILP.py and run the ILP.

License

This project is licensed under the MIT License. See the LICENSE file for details.