Mixed integer programming formulations for the robust resource constrained project scheduling problem.
This project investigates the impact of different formulations on the solution quality of the robust resource constrained project scheduling problem (RCPSP).
Two mixed-integer programming (MIP) formulations are presented under the assumption of deterministic activity processing times. These are found under the rcpsp/formulations/perfect_knowledge directory. rcpsp_dis.py separates the time horizon to discrete intervals while rcpsp_con.py assumes a continuous horizon .
Following a minimax bilevel approach, we formulate the worst case makespan robust RCPSP using the pyomo.bilevel package. The bilevel models are reformulated to each of the perfect-knowledge MIP formulations and then solved using the CPLEX solver. These can be found under the rcpsp/formulations/robust directory.
For every formulation we implement a pyomo abstract model and a corresponding solver class that instantiates the model, computes a warm start solution using a serial schedule generation scheme rcpsp/heuristics/sgs.py and invokes CPLEX using the SolverFactory wrapper.
The models are executed on the PSPLIB instances found under data/instances/json.
The project is structured as follows.
.
├── README.md
├── data
│ └── instances
│ ├── json
│ │ ├── j120
│ │ ├── j30
│ │ ├── j60
│ │ └── j90
│ └── sm
│ ├── j120
│ ├── j30
│ ├── j60
│ └── j90
└── rcpsp
├── formulations
│ ├── perfect_knowledge
│ │ ├── continuous
│ │ │ ├── rcpsp_con.py
│ │ │ └── solver_con.py
│ │ └── discrete
│ │ ├── rcpsp_dis.py
│ │ └── solver_dis.py
│ └── robust
│ ├── continuous
│ │ └── worst_case_makespan
│ │ ├── robust_rcpsp_con.py
│ │ └── robust_solver_con.py
│ └── discrete
│ └── worst_case_makespan
│ ├── robust_rcpsp_dis.py
│ └── robust_solver_dis.py
├── heuristics
│ ├── backward_recursion.py
│ ├── forward_recursion.py
│ └── sgs.py
└── util
├── instance_execution
│ ├── run_instances.sh
│ └── worst_case_regret.sh
├── instance_postprocessing
│ └── process_instance_results_extract_csv.py
├── instance_preprocessing
│ ├── convert_sm.sh
│ └── psplibconverter.py
└── priority_rule_evaluation
├── run_pr.sh
└── run_sgs.py