maxbannach/jdrasil

A Modular Library for Computing Tree Decompositions

Jdrasil

A Modular Library for Computing Tree Decompositions

Authors: Max Bannach, Sebastian Berndt, and Thorsten Ehlers

About

Jdrasil is a library to compute tree decompositions of simple, undirected graphs. It was developed for the first Parameterized Algorithms and Computational Experiments Challenge (PACE). It provides exact sequential and parallel, as well as heuristic, and approximation algorithms.

Jdrasil is build in a very modular fashion. This allows researchers to simply add new algorithms, heuristics, or preprocessing routines, which can then be combined in any way.

Installation

Jdrasil uses Gradle as build tool. Thanks to the gradle wrapper, nothing extra has to be installed in order to install Jdrasil.

To build Jdrasil, simply invoke the gradle build script:

cd Jdrasil
./gradlew assemble

There is also a bat-file for windows systems. After the script is finished, an executable jar will be placed in:

build/jars/Jdrasil.jar

The jar can be used as library or as standalone:

java −jar build/jars/Jdrasil.jar
java −cp build/jars/Jdrasil.jar jdrasil.Exact
java −cp build/jars/Jdrasil.jar jdrasil.Heuristic
java −cp build/jars/Jdrasil.jar jdrasil.Approximation

Building Start Scripts

Jdrasil comes with PACE like starting scripts: tw-exact, tw-heuristic, and tw-approximation. They can be build with gradle:

./gradlew exact
./gradlew heuristic
./gradlew approximation

Use the scripts as defined on the (PACE) website:

./tw−exact −s 42 < myGraph.gr > myGraph.td
./tw−heuristic −s 42 < myHugeGraph.gr > myHugeGraph.td

Build the Documentation

Jdrasil comes with a manual and JavaDocs. To build the manual, an up-to-date LuaLaTeX installation is required:

./gradlew manual
./gradlew javadoc

The manual will be placed in build/docs/manual/manual.pdf and the JavaDocs in build/docs/javadoc.

Build for PACE

Jdrasil provides a Gradle task that builds the PACE-version of Jdrasil. In particular, this will build Jdrasil and create the above mentioned start scripts.

./gradlew pace