main.py

#!/usr/bin/env python3

import logging
from argparse import ArgumentParser
from enum import Enum

from lot_sizing.model import MipModel, Input


class Linearisation(Enum):
    COMPACT = 1  # based on Liberti's linearisation
    EXTENDED = 2  # based on network flow formulation
    SIMPLE = 3  # based on Glover-Woolsey linearisation


module_logger = logging.getLogger('main')

cmd_parser = ArgumentParser(
    description='Integer programming formulations for the discrete, single-machine, multi-item, single-level lot sizing problem.')
cmd_parser.add_argument('-f', '--file', metavar="input_file", type=str,
                        help='File containing the input data',
                        required=True)
cmd_parser.add_argument('--lin', type=int, choices=range(1, 4), required=True,
                        help='Specifies which linearisation should be used: 1-compact, 2-extended, 3-glover_woolsey')
cmd_parser.add_argument('-s', '--solver', type=str, default='SCIP',
                        help='Specifies the solver used by ortools. Default is SCIP.')

if __name__ == '__main__':
    logging.basicConfig(level=logging.INFO)
    cmd_args = cmd_parser.parse_args()
    prob_input = Input.read_file(cmd_args.file)
    linearisation = Linearisation(cmd_args.lin)
    if linearisation == Linearisation.COMPACT:
        mip_model = MipModel.build_liberti_formulation(prob_input, cmd_args.solver)
    elif linearisation == Linearisation.EXTENDED:
        mip_model = MipModel.build_network_flow_formulation(prob_input, cmd_args.solver)
    elif linearisation == Linearisation.SIMPLE:
        mip_model = MipModel.build_glover_woolsey_formulation(prob_input, cmd_args.solver)
    else:
        raise RuntimeError('Unexpected linearisation.')
    schedule, cost = mip_model.compute_optimal_schedule()
    module_logger.info(f'Computed schedule: {schedule}')
    module_logger.info(f'Objective value: {cost}')