A Constraint Solver
OCaml C++ Standard ML Other
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
backend
frontend
.gitignore
LICENSE
OMakefile
OMakefile.config.sample
OMakeroot
README.md

README.md

Inez - A Constraint Solver

Introduction

Inez is a constraint solver.

Inez implements the ILP Modulo Theories scheme, as described in a CAV paper. Simply put, we combine a Mathematical Programming solver with background solvers.

Inez is OCaml-centric. The preferred mode of interacting with the solver is via scripts written in a Camlp4-powered superset of OCaml.

Inez is a research prototype, and may contain serious bugs. You should not use it in production.

Dependencies

Unix

Inez is known to work on x86_64 GNU/Linux and Mac OS X. Other modern Unixen should work, as long as all dependencies are satisfied.

Inez does not work on Windows. We depend quite heavily on Jane Street Core, which is Unix-only.

SCIP

Inez depends on the SCIP optimization suite, version 3.1.x. The SCIP optimization suite is available without charge for academic and non-commercial purposes. For other purposes, a license agreement is required.

Once you obtain the "optimization suite" distribution, the following recipe suffices (applied to the toplevel directory):

make scipoptlib \
    SHARED=true \
    READLINE=false \
    ZLIB=false \
    GMP=false \
    ZIMPL=false

The .so is under lib/. Create a symbolic link to it as follows:

ln -s libscipopt-3.1.0.linux.x86_64.gnu.opt.so libscipopt.so

C and C++ Libraries and Tools

Building Inez requires GCC (C++ frontend included). We also depend on Boost. Fetch and untar a fresh tarball.

OCaml Libraries and Tools

Inez requires OCaml 4.02 or newer. Additionally, you will need recent versions of the following packages:

  • async
  • camlidl
  • camlp4
  • comparelib
  • core
  • herelib
  • ocamlfind
  • ocaml_plugin
  • omake
  • sexplib

The best way to obtain the above is via OPAM.

Installation

Once all dependencies are satisfied, you have to:

  • Copy OMakefile.config.sample to OMakefile.config.
  • Adapt OMakefile.config for your setup.
  • Type omake frontend/inez.opt to build Inez.

inez.opt is dynamically linked against libscipopt.so. Modify LD_LIBRARY_PATH so that the dynamic linker can find this shared library. On OS X, you should set DYLD_FALLBACK_LIBRARY_PATH in place of LD_LIBRARY_PATH.

inez.opt accepts a single argument for the input program. Some examples can be found under frontend/examples.