Bayes Blocks libraryversion 1.1.1, 2007-01-03SYSTEM REQUIREMENTS --------------------------------------------------  Python toolboxThe Python version of the library should in principle work under anyplatform Python supports that has a standards compliant C++ compiler.Specific software requirements are as follows:- Python 2.3 or later (earlier versions may work, but are unsupported)- NumPy or Numerical PythonOptional software:- MATLAB 6.5 or later (for saving and loading .mat files)- Graphviz 1.9 or later (for BBVis visualisation package)- fig2dev from TransFig (for BBVis visualisation package)- Python Megawidgets 1.1 or later (for BBVis visualisation package)- SWIG 1.3.17 or later (for recompiling from scratch)Most of these are included in modern Linux distributions.  Pre-builtWindows binaries are provided at the web page of the library, but theydo not include support for Matlab integration.The files compiled by SWIG are included in the distribution so SWIG isnot needed to compile the distributed version of the library, but onlyif some modifications are made.  Standalone C++ libraryThe underlying library is primarily designed for and should work underany reasonably modern Un*x operating system.  We have used variousversions of Linux, Mac OS X, IRIX 6.5, HP Tru64 Unix and Solaris.Others should work but may require some fiddling with compileroptions.  Some platform specific hints for compilation are givenbelow.For using only the C++ library, only additional software required area standards compliant C++ compiler and GNU make.COMPILATION ----------------------------------------------------------  Python toolboxTo compile the Python version, use  python setup.py build  python setup.py installor other standard Python setup.py magic.  The setup script tries todetect a Matlab installation to compile the library with Matlabsupport (i.e. ability to save the results in .mat files).  In order tochange the name of Matlab executable from the default 'matlab', pleaseedit the file setup.cfg.  Standalone C++ libraryThe package contains a configure script to detect compiler andother settings so the plain C++ library can be compiled by thestandard  ./configure  make method.  The configure script can take options to specifywhether Matlab is used.  Run it with option '--help' for moreinformation.PLATFORM SPECIFIC INSTRUCTIONS ---------------------------------------  WindowsA Python version of the library should compile using setup.py withMicrosoft Visual Studio C++.  As the library has been developed mainlyunder Linux it should compile with GCC derivatives (MinGW, DJGPP), butcompiling Python extensions with those is in general a nontrivialtask.  Provided setup.py disables Matlab hooks under Windows, but itmay be possible to restore them by finding suitable compiler flagsfor include and library paths.  Mac OS XCompilation of the Python library with Matlab binding yields someerrors at the end, but they can be safely ignored.  To use the Pythonlibrary with Matlab bindings, you must set the value of theenvironment variable DYLD_LIBRARY_PATH to <Matlab root>/bin/mac,e.g. DYLD_LIBRARY_PATH=/Applications/MATLAB704/bin/mac.  IRIX 6.5Compilation of the C++ module for Python may be tricky.  In case ofproblems, try upgrading to the latest Python release and apply thefollowing patch to Python:http://sourceforge.net/tracker/?func=detail&aid=1222585&group_id=5470&atid=305470  Tru64 UnixCompilation of the C++ module for Python may be tricky.  In case ofproblems, try upgrading to the latest Python release and apply thefollowing patch to Python:http://sourceforge.net/tracker/?func=detail&aid=1222585&group_id=5470&atid=305470HP compiler seems to require additional flags to use standardscompliant libraries.  This can be addressed by setting environmentvariable CXX to value 'cxx -D__USE_STD_IOSTREAM' before running'python setup.py build'.USAGE ----------------------------------------------------------------Please see the examples.py available together with the full package.For realistic examples of using the library, there are extra packageswith code used in the experiments of our papers available at the webpage  http://www.cis.hut.fi/projects/bayes/Some documentation for the package is included in subdirectory 'doc'of the main archive.REFERENCE ------------------------------------------------------------If you use the software in your work, please cite:  M. Harva, T. Raiko, A. Honkela, H. Valpola, and J. Karhunen.  Bayes Blocks: An Implementation of the Variational Bayesian  Building Blocks Framework. In Proceedings of the 21st Conference  on Uncertainty in Artificial Intelligence, Edinburgh, UK,  pp. 259-266, 2005.LEGALESE -------------------------------------------------------------Copyright (C) 2001-2006 by Markus Harva, Antti Honkela, AlexanderIlin, Tapani Raiko, Harri Valpola and Tomas 謘tman.This program is free software; you can redistribute it and/ormodify it under the terms of the GNU General Public Licenseas published by the Free Software Foundation; either version 2of the License, or (at your option) any later version.This program is distributed in the hope that it will be useful,but WITHOUT ANY WARRANTY; without even the implied warranty ofMERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See theGNU General Public License for more details.You should have received a copy of the GNU General Public Licensealong with this program; if not, write to the Free SoftwareFoundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.CONTACT INFORMATION --------------------------------------------------The WWW home page of this library is:  http://www.cis.hut.fi/projects/bayes/The authors can be reached by paper mail:  Bayes group  Laboratory of Computer and Information Science  P.O. Box 5400  FI-02015 TKK  Finlandor by email:  bayeslib@cis.hut.fi