Usage

To get started, you need to install the package. You can do this by running the following command:

pip install jsp-instance-utils

Then you can import instances from the package and use them in your code. For example, to load an instance from the package. Here is an example of how to load the ft06 instance from the package:

from jsp_instance_utils.instances import ft06
from jsp_instance_utils.instances import ft06_makespan
from jsp_instance_utils.instances import ft06_makespan_is_optimal

or you can do it in a more compact way:

from jsp_instance_utils.instances import ft06, ft06_makespan, ft06_makespan_is_optimal

The individual instances follow the naming in the Table of isntances. The _makespan-suffix indicates yields the lowest makespan of the instance that is known (cf. the cited literature). For small instances this is the optimal makespan, for larger instances it is the best known makespan. The _makespan_is_optimal-suffix indicates whether the _makespan-value is optimal or not. So ft06 refers to the instance (in the format of this software package), ft06_makespan to the best known makespan, and ft06_makespan_is_optimal to whether the best known makespan is optimal or not. The ft06 instance is a small instance with 6 jobs and 6 machines and was introduced by Fisher and Thompson in 1963.

Here is another example of how to load the ta01 instance from the package:

from jsp_instance_utils.instances import ta01, ta01_makespan, ta01_makespan_is_optimal

The ta01 instance is a instance with 10 jobs and 10 machines and was introduced by Taillard in 1993.

Format of the Instances

Note

The machines are indexed starting from 0.

The instances in this package are represented as numpy arrays with shape (n_jobs, n_machines, 2). here is a small example:

import numpy as np
custom_jsp_instance = np.array([
    [
        [0, 1, 2, 3],  # job 0 (machine0, machine1, machine2, machine3)
        [0, 2, 1, 3]  # job 1 (machine0, machine1, machine2, machine3)
    ],
    [
        [11, 3, 3, 12],  # task durations of job 0
        [5, 16, 7, 4]  # task durations of job 1
    ]
])

This format is useful, because the individual matrices can easily be extracted. Here is an example:

machine_matrix, processing_time_matrix = custom_jsp_instance

That way machine_matrix would be:

np.array([
    [0, 1, 2, 3],
    [0, 2, 1, 3]
])

and processing_time_matrix would be:

np.array([
    [11, 3, 3, 12],
    [5, 16, 7, 4]
])

Here is a complete runnable example of how to load the ft06 instance from the package and how to extract the machine matrix and the processing time matrix:

from jsp_instance_utils.instances import ft06, ft06_makespan, ft06_makespan_is_optimal

machine_matrix, processing_times = ft06

print(f"""
The machine matrix of instance 'ft06' is:
{machine_matrix}

The processing times of instance 'ft06' are:
{processing_times}

The lowest known makespan of instance 'ft06' is:
{ft06_makespan}
The lowest known makespan is a {'optimal' if ft06_makespan_is_optimal else 'not necessarily optimal'} solution.
""")
    

This code snippet returns the following output:

Analogously, you can load the ta01 instance from the package and extract the machine matrix and the processing time matrix:

from jsp_instance_utils.instances import ta01, ta01_makespan, ta01_makespan_is_optimal

machine_matrix, processing_times = ta01

print(f"""
The machine matrix of instance 'ft06' is:
{machine_matrix}

The processing times of instance 'ft06' are:
{processing_times}

The lowest known makespan of instance 'ft06' is:
{ta01_makespan}
The lowest known makespan is a {'optimal' if ta01_makespan_is_optimal else 'not necessarily optimal'} solution.
""")

This code snippet yields the following output:

Reinforcement Learning Environments

For the usage of the instances in reinforcement learning environments, we provide a wrapper class that can be used to create a reinforcement learning environment please check out the Table of Instances of this project.