Linear Optimization Algorithm: Exact vs. Metaheuristic Comparative Study
Budget: $250 – $350 USD
For my current academic study I need a complete optimisation workflow that I can later translate into a journal-ready case study. The work is split into three tightly-connected pieces:
1. Build a full linear programming model in Gurobi. Please set out decision variables, objective function and all constraints clearly in code and in a short accompanying write-up. The problem will be solved on a publicly available dataset so every result can be reproduced and cited.
2. Develop a stand-alone metaheuristic—Genetic Algorithm, Simulated Annealing, or another well-defended choice. I’m happy with a clean Python implementation that uses standard scientific libraries (NumPy, pandas, matplotlib) so I can run repeated experiments on my own machine.
3. Run an experimental comparison between the exact Gurobi solution and the metaheuristic across multiple random seeds. I need solution quality statistics, computation times, convergence plots and a brief discussion of trade-offs. All figures should be exported as high-resolution images suitable for a paper.
Deliverables
• Annotated Gurobi model script or Jupyter notebook
• Metaheuristic source code with a runnable main file
• Reproducibility package (dataset link, parameter files, README)
• Short report (3-5 pages) summarising methodology, results, plots and key insights
Acceptance criteria
• Gurobi model returns an optimality gap of 0.0 on the dataset provided
• Metaheuristic reaches within an agreed percentage of the optimal objective in at least 80 % of runs
• All code executes without manual edits on a fresh environment (Python ≥3.9, Gurobi ≥10)
Only freelancers with proven experience in advanced Gurobi modelling and academic-grade metaheuristic design should engage; please reference at least one previous optimisation project in your bid.
1. Build a full linear programming model in Gurobi. Please set out decision variables, objective function and all constraints clearly in code and in a short accompanying write-up. The problem will be solved on a publicly available dataset so every result can be reproduced and cited.
2. Develop a stand-alone metaheuristic—Genetic Algorithm, Simulated Annealing, or another well-defended choice. I’m happy with a clean Python implementation that uses standard scientific libraries (NumPy, pandas, matplotlib) so I can run repeated experiments on my own machine.
3. Run an experimental comparison between the exact Gurobi solution and the metaheuristic across multiple random seeds. I need solution quality statistics, computation times, convergence plots and a brief discussion of trade-offs. All figures should be exported as high-resolution images suitable for a paper.
Deliverables
• Annotated Gurobi model script or Jupyter notebook
• Metaheuristic source code with a runnable main file
• Reproducibility package (dataset link, parameter files, README)
• Short report (3-5 pages) summarising methodology, results, plots and key insights
Acceptance criteria
• Gurobi model returns an optimality gap of 0.0 on the dataset provided
• Metaheuristic reaches within an agreed percentage of the optimal objective in at least 80 % of runs
• All code executes without manual edits on a fresh environment (Python ≥3.9, Gurobi ≥10)
Only freelancers with proven experience in advanced Gurobi modelling and academic-grade metaheuristic design should engage; please reference at least one previous optimisation project in your bid.