JBenge - Benchmark Generator

We provide a benchmark generator for normal logic programs, which can be used for answer set programming systems.

The benchmark generator is implemented in Java, under JDK 1.4.2. The different encodings for these problems are taken from smodels, DLV, J. Ward, J. Schlipf, F. Lin, J. Zhao, and TheoryBase.

We provide two Java archive files, one with a graphical interface and another one, which can directly be used on the shell.

Download jbenge.jar [Version: June 14, 2006], benchmark generator with graphical interface
Download benchmarks.jar [Version: June 14, 2006] , pure benchmark generator

Download the jar-File
Start the application by
```
java -jar jbenge.jar
```
A Dialog window is opened
Follow the Instructions

Call:
java -jar benchmarks.jar [-graph=Graph -n=NBG] [-encoding=Enc [-nc=NBE]] [-planning=Plan [-np=NBP]] [-o=OutFile]
-graph=Graph -n=NBG specifies a graph including parameter.
-encoding=Enc [-nc=NBE] specifies an encoding
-planning=Plan [-np=NBP] specifies a planning problem
-o=OutFile specifies an output file

Detailed command options are given in the tables below.

Graph Problems

We have the following possibilities for generating graph instances:

Graph Problem	Option in JBenge	required Parameter	Call for benchmarks.jar
Circle graph with n vertices	circle	n	-graph=circle -n=n,0,0
Complete graph with n vertices	complete	n	-graph=complete -n=n,0,0
Ladder graph with n runges	ladder	n	-graph=ladder -n=n,0,0
Simplex graph with n levels	simplex	n	-graph=simplex -n=n,0,0
K _n×m graph with n upper and m lower vertices	knm	n and m	-graph=knm -n=n,m,0
Random graph with n vertices, where an edge between two vertices has propability p	random	n and p	-graph=random -n=n,0,p
Windmill graph with n wings	windmill	n	-graph=windmill -n=n,0,0
Clumpy graph with n clumps, where each clump has m vertices	clumpy	n and m	-graph=clumpy -n=n,m,0
Smallworld graph with propability p	smallworld	n,m, and p	-graph=smallworld -n=n,m,p

The graphs are presented as facts

			vtx(X). 
			edge(X,Y).

for every vertex X and every edge from X to Y. Undirected graphs contain

 edge(X,Y). edge(Y,X).

for every edge between X and Y.

These graph instances can be connected with one of the following encodings:

Encoding	Option in JBenge	required Parameters	requires a graph	Call for benchmarks.jar
Independent set problem	indset	-	yes	-encoding=indset
Kernel Problem	kernel	-	yes	-encoding=kernel
Coloring Problem	coloring	number of colors	yes	-encoding=coloring -nc=nbcolors
Hamiltonian cycle problem (by I. Niemelä) (Hint!)	hamiltonN	-	no	-encoding=hamiltonN
Hamiltonian cycle problem (by Lin and Zhao) (Hint!)	hamiltonL	-	no	-encoding=hamiltonL
Hamiltonian cycle problem (by Ward ans Schlipf) (Hint!)	hamiltonWS	-	no	-encoding=hamiltonWS

The option „no-encoding“ generates only the instance of a graph, whenever a graph has been choosen.

Important Hint for Hamilton Encoding: For the Hamilton encodings the definition of the initial vertex must be done separately. That is, you have to add the rule

		 initialvtx(x).

to the program, where "x" denotes the initial node for the Hamiltonian cycle.

Planning problems

Considering planning problems and other program instances, we provide the following encodings:

Planning Problem	Option in JBenge	required Parameter	Call for benchmarks.jar
Queens problem for a chess board of size n*n	queens	n	-planning=queens -np=n
Schur problem of partitioning n integers into m boxes	schur	n and m	-planning=schur -np=n,m
Logic program from nomore++	headn	n	-planning=headn -np=n
Logic program from nomore++	bodyn	n	-planning=bodyn -np=n

June 14, 2006: Deletion of syntactical errors in Hamiltonian Encodings

The tool is still under development. If you

detect some errors, or

have suggestions for including further options,

please contact konczak@cs.uni-potsdam.de.

JBenge: Benchmark Generator for answer set programming

Introduction

How to use jbenge.jar

How to use benchmarks.jar

Options

Graph Problems

Planning problems

History

Any Comments?