1 libDAI - A free/open source C++ library for Discrete Approximate Inference methods

2 ==================================================================================

4 v 0.2.1 - May 26, 2008

7 Copyright (C) 2006-2008 Joris Mooij [j dot mooij at science dot ru dot nl]

8 Radboud University Nijmegen, The Netherlands

11 ----------------------------------------------------------------------------------

12 This file is part of libDAI.

14 libDAI is free software; you can redistribute it and/or modify

15 it under the terms of the GNU General Public License as published by

16 the Free Software Foundation; either version 2 of the License, or

17 (at your option) any later version.

19 libDAI is distributed in the hope that it will be useful,

20 but WITHOUT ANY WARRANTY; without even the implied warranty of

21 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the

22 GNU General Public License for more details.

24 You should have received a copy of the GNU General Public License

25 along with libDAI; if not, write to the Free Software

26 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA

27 ----------------------------------------------------------------------------------

31 What is libDAI?

32 ---------------

33 libDAI is a free/open source C++ library (licensed under GPL, see the file

34 COPYING for more details) that provides implementations of various

35 (deterministic) approximate inference methods for discrete graphical models.

36 libDAI supports arbitrary factor graphs with discrete variables (this includes

37 discrete Markov Random Fields and Bayesian Networks).

39 libDAI is not intended to be a complete package for approximate inference.

40 Instead, it should be considered as an "inference engine", providing various

41 inference methods. In particular, it contains no GUI, currently only supports

42 its own file format for input and output (although support for standard file

43 formats may be added), and provides no visualization.

45 Because libDAI is implemented in C++, it is very fast compared with e.g. MatLab

46 implementations. libDAI does provide a MatLab interface for easy integration

47 with MatLab. Currently, libDAI supports the following deterministic approximate

48 inference methods:

50 * Mean Field

51 * (Loopy) Belief Propagation

52 * Tree Expectation Propagation

53 * Generalized Belief Propagation

54 * Double-loop GBP

55 * Loop Corrected Approximate Inference

57 Exact inference by JunctionTree is also provided.

59 Many of these algorithms are not yet available in similar open source software,

60 to the best of the author's knowledge (open source packages supporting both

61 directed and undirected graphical models are Murphy's BNT, Intel's PNL and gR).

63 The library is targeted at researchers; to be able to use the library, a good

64 understanding of graphical models is needed. However, the code will hopefully

65 find its way into real-world applications as well.

68 Rationale

69 ---------

70 In my opinion, the lack of open source reference implementations hampers

71 progress in research on approximate inference. Methods differ widely in terms

72 of quality and performance characteristics, which also depend in different ways

73 on various properties of the graphical models. Finding the best approximate

74 inference method for a particular application therefore often requires

75 empirical comparisons. However, implementing and debugging these methods takes

76 a lot of time which could otherwise be spent on research. I hope that this code

77 will aid researchers to be able to easily compare various (existing as well as

78 new) approximate inference methods, in this way accelerating research and

79 stimulating real-world applications of approximate inference.

82 Releases

83 --------

84 Releases can be obtained from http://www.mbfys.ru.nl/~jorism/libDAI.

85 License: GNU Public License v2 (or higher).

87 libDAI-0.2 December 1, 2006

88 libDAI-0.2.1 May 26, 2008

91 Acknowledgments

92 ---------------

93 The development reported here is part of the Interactive Collaborative

94 Information Systems (ICIS) project, supported by the Dutch Ministry of Economic

95 Affairs, grant BSIK03024. I would like to thank Martijn Leisink for providing

96 the basis on which libDAI has been built.

99 Known issues

100 ------------

101 Due to a bug in GCC 3.3.x and earlier (http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20358)

102 it doesn't compile with these versions (it does compile with GCC version 3.4 and higher).

103 Workaround: replace the two NAN's in factor.h causing the error messages by -1.

106 Documentation

107 -------------

108 Almost nonexistant. But I'm working on it. In the meantime, I'll provide limited support

109 by email. The following gives an overview of different methods and their properties

110 (can be slightly obsolete):

112 BP

113 updates UpdateType SEQFIX,SEQRND,SEQMAX,PARALL

114 tol double

115 maxiter size_t

116 verbose size_t

117 MF

118 tol double

119 maxiter size_t

120 verbose size_t

121 HAK

122 clusters MIN,DELTA,LOOP

123 loopdepth

124 doubleloop bool

125 tol double

126 maxiter size_t

127 verbose size_t

128 JTREE

129 updates UpdateType HUGIN,SHSH

130 verbose size_t

131 MR

132 updates UpdateType FULL,LINEAR

133 inits InitType RESPPROP,CLAMPING,EXACT

134 verbose size_t

135 TREEEP

136 type TypeType ORG,ALT

137 tol double

138 maxiter size_t

139 verbose size_t

140 LC

141 cavity CavityType FULL,PAIR,PAIR2,UNIFORM

142 updates UpdateType SEQFIX,SEQRND(,NONE)

143 reinit bool

144 cavainame string

145 cavaiopts Properties

146 tol double

147 maxiter size_t

148 verbose size_t

152 Quick start

153 -----------

154 You need:

155 - a recent version of gcc (version 3.4 at least)

156 - the Boost C++ libraries (under Debian/Ubuntu you can install them using

157 "apt-get install libboost-dev libboost-program-options-dev")

158 - GNU make

160 To build the source, edit the Makefile and then run

162 make

164 If the build was successful, you can test the example program:

166 ./example tests/alarm.fg

168 or the more elaborate test program:

170 tests/test --aliases tests/aliases.conf --filename tests/alarm.fg --methods JTREE_HUGIN BP_SEQMAX

172 A description of the factor graph (.fg) file format can be found in the file FILEFORMAT.