Demo OptaPlanner

Introduction
Documentation
Technology
Example
- Initialization
- Running
- Run
- Output

Introduction

This is a demonstration of OptaPlanner, which is a constraint satisfaction solver tool.

When the demo app runs, it creates a demo constraint satisfaction problem, then solves it.

The constraint satisfaction problem is how to match makers (a.k.a. producers) with takers (a.k.a. consumers).

Documentation

Documentation within this repo:

Documentation

Technology

The software implementation uses:

Java programming language
OptaPlanner open source contraint satisfaction solver
XStream XML serializer/deserializer
SLF4J (Simple Logging Facade for Java) for log messages.
Google Guava core libraries for Java, such as for collections.
Apache Commons for Lang and [Math].

Example

When the app runs, it creates a demo constraint satisfaction problem, then solves it.

The constraint satisfaction problem is how to match makers (a.k.a. producers) with takers (a.k.a. consumers).

Initialization

Initialization of facts:

Create 10 makers. These are named 0 to 9.
Create 10 takers. These are named 0 to 9.

Initialization of planning entities:

Create 10 matchers. Each matcher can associate a maker and taker.
Each matcher is loaded with a corresponding maker, because this represents the facts i.e. we want to match all the makers.
Each matcher has null for the taker, because this is the planning variable i.e. we want to try to match on various takers.

Running

Scorer pseudocode:

Score each matcher by comparing its maker and taker.
When the maker name and taker name are equal, then score +1 soft, else score -1 soft.
The scorer will maximize matchers that associate a maker and taker with the same name.

Solution pseudocode:

Create an OptaPlanner solver, and load it with the makers, takers, matchers, etc.
Create a timeout of 1 minute. This is fine for the demo small scale.
The solver will try associating the matchers with various takers, then report results.

See setup.md

Output

The output shows the solution successfully associates all makers and takers.

The score 0hard means the hard constraint score is 0, which is correct because the scorer had no hard contraints.

The score 10soft means the soft constraint score is 10, which is correct because the solver matched all 10 makers and takers.

Output with log level info:

Demo OptaPlanner version 1.0.0

Initialize...
    name:Demo Solution,
    score:(0hard/0soft),
    matchers:[
        matcher:{name:0,maker:{name:0},taker:{null}},
        matcher:{name:1,maker:{name:1},taker:{null}},
        matcher:{name:2,maker:{name:2},taker:{null}},
        matcher:{name:3,maker:{name:3},taker:{null}},
        matcher:{name:4,maker:{name:4},taker:{null}},
        matcher:{name:5,maker:{name:5},taker:{null}},
        matcher:{name:6,maker:{name:6},taker:{null}},
        matcher:{name:7,maker:{name:7},taker:{null}},
        matcher:{name:8,maker:{name:8},taker:{null}},
        matcher:{name:9,maker:{name:9},taker:{null}}]
Solved...
    name:Demo Solution,
    score:(0hard/10soft),
    matchers:[
        matcher:{name:0,maker:{name:0},taker:{name:0}},
        matcher:{name:1,maker:{name:1},taker:{name:1}},
        matcher:{name:2,maker:{name:2},taker:{name:2}},
        matcher:{name:3,maker:{name:3},taker:{name:3}},
        matcher:{name:4,maker:{name:4},taker:{name:4}},
        matcher:{name:5,maker:{name:5},taker:{name:5}},
        matcher:{name:6,maker:{name:6},taker:{name:6}},
        matcher:{name:7,maker:{name:7},taker:{name:7}},
        matcher:{name:8,maker:{name:8},taker:{name:8}},
        matcher:{name:9,maker:{name:9},taker:{name:9}}]

joelparkerhenderson / demo-optaplanner