OptaPy

OptaPy is an AI constraint solver for Python to optimize the Vehicle Routing Problem, Employee Rostering, Maintenance Scheduling, Task Assignment, School Timetabling, Cloud Optimization, Conference Scheduling, Job Shop Scheduling, Bin Packing and many more planning problems.

OptaPy wraps the OptaPlanner engine internally, but using OptaPy in Python is significantly slower than using OptaPlanner in Java or Kotlin.

Get started

Prerequisites

  1. Install Python 3.9 or later
  2. Install JDK 11 or later with JAVA_HOME configured appropriately.

Build

Install the python build module (if not already installed):

pip install build

In the optapy-core directory, use the command below to build the optapy python wheel into the dist directory:

cd optapy-core
python -m build

Install it into a virtual environment using pip:

# Activate a virtual environment first
source my_project/venv/bin/activate
pip install dist/optapy-0.0.0-py3-none-any.whl

Run

Running run.sh runs optapy-quickstarts/school-timetabling/main.py after building optapy and installing it to optapy-quickstarts/school-timetabling/venv.

Source code overview

Domain

In OptaPlanner, the domain has three parts:

  • Problem Facts, which do not change
  • Planning Entities, which have one or more planning variables
  • Planning Solution, which define the facts and entities of the problem

Problem Facts

To declare Problem Facts, use the @problem_fact decorator

from optapy import problem_fact
@problem_fact
class Timeslot:
    def __init__(self, id, dayOfWeek, startTime, endTime):
        self.id = id
        self.dayOfWeek = dayOfWeek
        self.startTime = startTime
        self.endTime = endTime

Planning Entities

To declare Planning Entities, use the @planning_entity decorator

from optapy import planning_entity, planning_id, planning_variable

@planning_entity
class Lesson:
    def __init__(self, id, subject, teacher, studentGroup, timeslot=None, room=None):
        self.id = id
        self.subject = subject
        self.teacher = teacher
        self.studentGroup = studentGroup
        self.timeslot = timeslot
        self.room = room

    @planning_id
    def getId(self):
        return self.id

    @planning_variable(Timeslot, value_range_provider_refs=["timeslotRange"])
    def getTimeslot(self):
        return self.timeslot

    def setTimeslot(self, newTimeslot):
        self.timeslot = newTimeslot

    @planning_variable(Room, value_range_provider_refs=["roomRange"])
    def getRoom(self):
        return self.room

    def setRoom(self, newRoom):
        self.room = newRoom
  • @planning_variable method decorators are used to indicate what fields can change. The method MUST follow JavaBean style conventions and have a corresponding setter (i.e. getRoom(self), setRoom(self, newRoom)). The first parameter of the decorator is the type of the Planning Variable (required). The value_range_provider_refs parameter tells OptaPlanner what value range providers on the Planning Solution this Planning Variable can take values from.
  • @planning_id is used to uniquely identify an entity object of a particular class. The same Planning Id can be used on entities of different classes, but the ids of all entities in the same class must be different.

Planning Solution

To declare the Planning Solution, use the @planning_solution decorator

from optapy import planning_solution, problem_fact_collection_property, value_range_provider, planning_entity_collection_property, planning_score

@planning_solution
class TimeTable:
    def __init__(self, timeslotList=[], roomList=[], lessonList=[], score=None):
        self.timeslotList = timeslotList
        self.roomList = roomList
        self.lessonList = lessonList
        self.score = score

    @problem_fact_collection_property(Timeslot)
    @value_range_provider(range_id = "timeslotRange")
    def getTimeslotList(self):
        return self.timeslotList

    @problem_fact_collection_property(Room)
    @value_range_provider(range_id = "roomRange")
    def getRoomList(self):
        return self.roomList

    @planning_entity_collection_property(Lesson)
    def getLessonList(self):
        return self.lessonList

    @planning_score(HardSoftScore)
    def getScore(self):
        return self.score

    def setScore(self, score):
        self.score = score
  • @value_range_provider(range_id) is used to indicate a method returns values a Planning Variable can take. It can be referenced by its id in the value_range_provider_refs parameter of @planning_variable. It should also have a @problem_fact_collection_property or a @planning_entity_collection_property.
  • @problem_fact_collection_property(type) is used to indicate a method returns Problem Facts. The first parameter of the decorator is the type of the Problem Fact Collection (required). It should be a list.
  • @planning_entity_collection_property(type) is used to indicate a method returns Planning Entities. The first parameter of the decorator is the type of the Planning Entity Collection (required). It should be a list.
  • @planning_score(scoreType) is used to tell OptaPlanner what field holds the score. The method MUST follow JavaBean style conventions and have a corresponding setter (i.e. getScore(self), setScore(self, score)). The first parameter of the decorator is the score type (required).

Constraints

You define your constraints by using the ConstraintFactory

import java
from domain import Lesson
from optapy import get_class, constraint_provider
from optapy.types import Joiners, HardSoftScore

# Get the Java class corresponding to the Lesson Python class
LessonClass = get_class(Lesson)

@constraint_provider
def defineConstraints(constraintFactory):
    return [
        # Hard constraints
        roomConflict(constraintFactory),
        # Other constraints here...
    ]

def roomConflict(constraintFactory):
    # A room can accommodate at most one lesson at the same time.
    return constraintFactory \
            .fromUniquePair(LessonClass, [
            # ... in the same timeslot ...
                Joiners.equal(lambda lesson: lesson.timeslot),
            # ... in the same room ...
                Joiners.equal(lambda lesson: lesson.room)]) \
            .penalize("Room conflict", HardSoftScore.ONE_HARD)

for more details on Constraint Streams, see https://docs.optaplanner.org/latest/optaplanner-docs/html_single/index.html#constraintStreams

| --- | --- |
| Note | Since from is a keyword in python, to use the constraintFactory.from function, you access it like constraintFactory.from_(class, [joiners…​]) |

Solve

from optapy import get_class, solve
from optapy.types import SolverConfig, Duration
from constraints import defineConstraints
from domain import TimeTable, Lesson, generateProblem
import java

solverConfig = SolverConfig().withEntityClasses(get_class(Lesson)) \
    .withSolutionClass(get_class(TimeTable)) \
    .withConstraintProviderClass(get_class(defineConstraints)) \
    .withTerminationSpentLimit(Duration.ofSeconds(30))

solution = solve(solverConfig, generateProblem())

solution will be a TimeTable instance with planning variables set to the final best solution found.

GitHub - optapy/optapy: OptaPlanner wrappers for Python. Currently significantly slower than OptaPlanner in Java or Kotlin.
OptaPlanner wrappers for Python. Currently significantly slower than OptaPlanner in Java or Kotlin. - GitHub - optapy/optapy: OptaPlanner wrappers for Python. Currently significantly slower than Op...
GitHuboptapy