How andreastocco01/or2 works
• Core Execution Module: Represented by main.c, which likely orchestrates the overall execution.
• TSP variants (tsp.c, tsp_greedy.c, tsp_diving.c, tsp_tabu.c, tsp_vns.c, tsp_localbranching.c, tsp_cplex.c)
• Mincut (mincut.c) and other optimization related modules like eventlog.c, chrono.c, util.c.
• Experiment and Results Management: The “runs” directory, where output files (for exact, mathematical, and other experiments) are stored.
• Tools: The “tools” directory (Python scripts like experiment.py, finalrun.sh, plot.py, etc.) which handle running experiments, profiling, and processing results.
• Configuration & Meta Information: The README and Makefile that guide experiment configurations (with hyperparameter settings indicated in the README e.g., config 0, 1, etc.).
Relationships and Data Flow
• Algorithm modules to the “Utility” components (chrono, eventlog, and util) which provide common functionality.
• That these modules output experiment data (results, logs) into the “runs” directory.
• That the “tools” (Python scripts) process the data from “runs” to generate plots, perform profiling, and possibly refine experiments.
Architectural Patterns and Design Principles
• Modularity: Each heuristic and algorithm is implemented in a separate module which not only supports comparative experimentation (as suggested by the hyperparameter tuning configurations) but also facilitates experimenting with different approaches.
• Separation of Concerns: The separation between the computational core (C modules for algorithms), experiment orchestration/analysis tools (Python scripts), and documentation (LaTeX files) should be clearly exposed.
• Experiment Management Flow: Include the “configuration” aspect (tuning parameters provided in the README) and illustrate how different configurations map to different experimental outcomes. This side-by-side comparison is central to the thesis research.
Specific Technologies and Tools
• That the project uses C for implementing heuristics and other computation-heavy components.
• Python is used as a support tool for experiments and analysis.
• LaTeX is an integral part of the reporting/documentation process.
• There is integration with CPLEX (as seen in tsp_cplex.c/h) indicating that some exact methods rely on external optimization libraries.