Skip to content
forked from emina/kodkod

A fork of Kodkod solver with Z3 Theorem Prover back-end for Relational Logic

License

Notifications You must be signed in to change notification settings

ModelWriter/kodkodz

 
 

Repository files navigation

Kodkod

This repository includes the source code for the Kodkod solver for relational logic. Kodkod provides a clean Java API for constructing, manipulating, and solving relational constraints. The source code is extensively documented, and the repository includes many examples demonstrating the use of the Kodkod API.

Kodkod is open-source and available under the MIT license. However, the implementation relies on third-party SAT solvers (SAT4J, MiniSat, Glucose, and (P)Lingeling), some of which are released under stricter licenses. Please see the solver licenses for details.

Downloading Kodkod

The easiest way to get started is to download the latest release, which includes precompiled binaries for Kodkod and various SAT solvers. You will need Java 8 running on Linux or Mac OS X.

Building Kodkod

The following instructions for building Kodkod from source have been tested on Linux (Fedora 22 with gcc 5.1.1) and on Mac OS X (10.10.5 with clang 6.0 and gcc 4.9). You may need to modify the build scripts for other operating systems.

Kodkod uses the Waf build system, which requires Python 2.5 or later. You will also need Java 8 and a C/C++ compiler, and your JAVA_HOME environment variable needs to point to the JDK 8 home directory.

  • Set the JAVA_HOME variable. For example, on OS X:

    $ export JAVA_HOME=`/usr/libexec/java_home`

  • Clone the kodkod repository:

    $ git clone https://github.com/emina/kodkod.git
    $ cd kodkod

  • Download Waf 1.8.12 and make it executable:

    $ wget --no-check-certificate https://waf.io/waf-1.8.12
    $ chmod u+x waf-1.8.12
    $ alias waf=$PWD/waf-1.8.12

  • Build the native libraries, kodkod.jar, and examples.jar and install them into the kodkod/lib directory:

    $ waf configure --prefix=. --libdir=lib build install

Running Kodkod

Download or build the kodkod.jar binary, solver binaries, and the examples.jar binary. Assuming that the current working directory contains these binaries, run the Sudoku example as follows:

$ java -cp kodkod.jar:examples.jar -Djava.library.path=. kodkod.examples.sudoku.Sudoku

The program will produce a solution to a sample Sudoku puzzle:

p cnf 3452 7954
primary variables: 486
translation time: 176 ms
solving time: 2 ms
+-------+-------+-------+
| 6 4 7 | 2 1 3 | 9 5 8 |
| 9 1 8 | 5 6 4 | 7 2 3 |
| 2 5 3 | 8 7 9 | 4 6 1 |
+-------+-------+-------+
| 1 9 5 | 6 4 7 | 8 3 2 |
| 4 8 2 | 3 5 1 | 6 7 9 |
| 7 3 6 | 9 2 8 | 1 4 5 |
+-------+-------+-------+
| 5 7 4 | 1 9 2 | 3 8 6 |
| 8 2 9 | 7 3 6 | 5 1 4 |
| 3 6 1 | 4 8 5 | 2 9 7 |
+-------+-------+-------+

About

A fork of Kodkod solver with Z3 Theorem Prover back-end for Relational Logic

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Java 79.2%
  • SMT 19.0%
  • Other 1.8%