You can find definitions of the job shop scheduling problem and the disjunctive MILP model in the course slides and/or textbook. In this section, we provide details on how we represent the problem instances and the conventions used.
Refer to the Coding Assignment #1 for the details.
The instances that are part of the solution to Coding Assignment #1 (see Quercus) are all valid instances for this assignment – however, the solutions to those instances for Coding Assignment #1 are not necessarily correct solutions to this assignment.[1] These are not the test instances that will be used on MarkUs so it is recommended that you also develop your own instances to fully test your code.
.
1.1 Function Definition
You must submit a Python file called coding2.py that contains (perhaps among other functions) a function with the following signature:
run disjunctive(instance), where instance is a string of the filename of the instance (e.g., “ft06.txt”).
The function should read in the problem instance (assume it is in the same folder as the coding2.py file), create a MILP model, send the model to Gurobi to find a solution, and interpret the results from Gurobi.
The function should return a tuple containing (this is the same format as the solution in Coding Assignment #1):
the makespan: integer
a schedule: a dictionary with keys corresponding to the job number and values being a list of start times for that job’s operations following that job’s operation order.
MarkUs will call your function and parse the return value to compare it against the correct answer. It is important that you follow the above instructions exactly.
1.2 Using the Gurobi Library and Python Interface
Gurobi and its Python interface are available on the ECF system that you all have remote access to. You can also download the library for academic use.
For all details about installing and examples files of using the interface, please see the Quercus announcement titled “Heads Up: Software to Use for Coding Assignment #2”.
