SPNBOX - A Toolbox for the Supervisory Control of Petri Nets

SPNBOX is a supervisory control toolbox for programs that can execute Matlab scripts (Matlab, Octave, FreeMat, SciLab, and others). There is also a C version of the toolbox that is included in the ACTS software,

The toolbox includes functions for the design of supervisors subject to liveness constraints and linear inequality constraints. A description of the functions included in the toolbox can be found in the manual. The theory of the implemented methods may be found in

Iordache M. V. and P. J. Antsaklis (2006), Supervisory Control of Concurrent Systems: A Petri Net Structural Approach, Birkhauser Boston.

The current version of the toolbox is 1.2 (July 2013). While the most recent version of the toolbox has been tested only in Octave, the toolbox is expected to work also with the other programs that can execute Matlab scripts, such as Matlab, FreeMat, and SciLab. Note that Octave, FreeMat, and SciLab are free software.

Requirements:

• One of Matlab, Octave, FreeMat, or SciLab. While the most recent version of the toolbox has been tested only in Octave, the toolbox is expected to work also with the other programs.
• The LPSOLVE software for mixed integer linear programming. LPSOLVE is free software. The toolbox has been tested with the version 5.5.2.0 of LPSOLVE.

Installing the toolbox:

• Download SPNBOX 1.2.
• Refer to the main directory of SPNBOX.
• If using Octave, copy ip_sc_octave.m to ip_sc.m. For other programs, copy the corresponding ip_sc_... file to ip_sc.m.
• If using an initialization script (such as startup.m) add the path of the SPNBOX files to the path list. You may do this with the addpath command.
• Install the driver file of LPSOLVE.
• Install the library files of LPSOLVE.

Installing the LPSOLVE driver: LPSOLVE is third-party software for mixed integer linear programming. The following appears to be a quick installation procedure. (This procedure was tested for Octave under Cygwin. Detailed information about the installation of the LPSOLVE driver may be found in the LPSOLVE documentation.)

• To check whether LPSOLVE is already installed, run iptst.m from the examples folder of SPNBOX. If the program completes successfully, SPNBOX is ready to use and no further action is necessary.
• If running the programs directly in Windows, not via Cygwin (for Cygwin, see the Unix procedure):
  • Download the appropriate driver files from http://sourceforge.net/projects/lpsolve/files/lpsolve/. In the version 5.5.2.0 of the package the driver archive for Matlab and 64 bit Windows is called lp_solve_5.5.2.0_MATLAB_exe_win64.zip, the driver for Octave and 32 bit Windows is called lp_solve_5.5.2.0_octave_exe_win32.zip, and so on.
  • Extract the zip archive and identify the driver file (octlpsolve.oct for Octave).
  • Copy the driver file to a directory on the path of Matlab/Octave/FreeMat/SciLab. (If unsure, type path in Matlab/Octave/FreeMat/SciLab to display the directories on their path.)
• If running the programs in Unix (Linux, Cygwin, ...) it might be best to build the driver from the source files:
  • To build the driver from the source, the program mkoctfile is needed for Octave (it might be found in the headers package of Octave) and mex for Matlab.
  • Download the zip file spnbox_1.2_third_party.zip. (This file merges the archives lp_solve_5.5.2.0_source.tar, lp_solve_5.5.2.0_MATLAB_source.tar, lp_solve_5.5.2.0_FreeMat_source.tar, lp_solve_5.5.2.0_octave_source.tar, and lp_solve_5.5.2.0_scilab_source.tar from http://sourceforge.net/projects/lpsolve/files/lpsolve/5.5.2.0/. This merge ensures that all necessary files are found when building the binary files.)
  • Extract the zip archive.
  • The path to the driver directories is third_party/lp_solve_5.5/extra/.
  • Go to the appropriate directory (such as octave/lpsolve for Octave).
  • Run sh ccc in a terminal window (for Octave and FreeMat) or Makefile.m in Matlab or builder.sce in SciLab.
  • Continue with the next step ONLY if the previous step was completed without errors.
  • Identify the driver file (octlpsolve.oct for Octave).
  • If the driver file is not already on the path of Matlab/FreeMat/Octave/SciLab, place it on that path. (If unsure, type path in Matlab/FreeMat/Octave/SciLab to display the directories on the path.)

Installing the library files of LPSOLVE: LPSOLVE is third-party software for mixed integer linear programming. The following appears to be a quick installation procedure. (This procedure was tested in Cygwin. Detailed information about the installation of the LPSOLVE library files may be found in the LPSOLVE documentation.)

• Once the LPSOLVE driver has been installed (see above), it is possible to check whether the library files are already installed by running iptst.m from the examples folder of SPNBOX. If the program completes successfully, SPNBOX is ready to use and no further action is necessary.
• If running the programs directly in Windows, not via Cygwin (for Cygwin, see the Unix procedure):
  • Download the library files from http://sourceforge.net/projects/lpsolve/files/lpsolve/. The archive file might be called lp_solve_5.5.2.0_dev_win64.zip for 64 bit Windows and lp_solve_5.5.2.0_dev_win32.zip for 32 bit Windows.
  • Extract the archive and copy the library files to a folder in the path list (see the PATH environment variable) such as windows\system32.
• If running the programs in Unix (Linux, Cygwin, ...) it might be best to build the library files from the source files:
  • The source files of the version 5.5.2.0 may be found in the archive spnbox_1.2_third_party.zip. For other versions see http://sourceforge.net/projects/lpsolve/files/lpsolve/.
  • Extract the zip archive.
  • Go to the lpsolve55 directory.
  • Run sh ccc in a terminal window.
  • This should create a bin directory and place the library files in it. For 32 bit Unix the files are located in bin/ux32/ are named liblpsolve55.a and liblpsolve55.so.
  • Copy the library files to a directory on the library path, such as /lib or /usr/lib.

Old Versions of SPNBOX

The first version of the toolbox can be used in Matlab only. It was developed with Matlab 5.3 and 6.0 under SunOS and LPSOLVE 3.2. To install the first version of the toolbox, follow this procedure:

• Download spnbox.zip.
• Extract the files of spnbox.zip and place them in some directory; from now on, let's call that directory Spnbox.
• Download the free software package LPSOLVE, version 3.2, or any other compatible version.
• Let's call Lpsolve the directory in which the source files of LPSOLVE have been placed.
• Download auxfiles.zip.
• Extract the files of auxfiles.zip and place them in the directory Lpsolve.
• In Matlab, change the directory to Lpsolve and then run lp2mex.
• Move the ipslv.mexsol... file from Lpsolve to the directory Spnbox