Cleanup of BBP code
[libdai.git] / include / dai / doc.h
1 /* Copyright (C) 2006-2008 Joris Mooij [joris dot mooij at tuebingen dot mpg dot de]
2 Radboud University Nijmegen, The Netherlands /
3 Max Planck Institute for Biological Cybernetics, Germany
4
5 This file is part of libDAI.
6
7 libDAI is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2 of the License, or
10 (at your option) any later version.
11
12 libDAI is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with libDAI; if not, write to the Free Software
19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22
23 /** \file
24 * \brief Contains additional doxygen documentation
25 *
26 * \todo Merge COPYING into doxygen documentation
27 * \todo Merge README into doxygen documentation
28 * \todo Document examples, tests and utils
29 *
30 * \todo Adapt (part of the) guidelines in http://www.boost.org/development/requirements.html#Design_and_Programming
31 *
32 * \todo Use "gcc -MM" to generate dependencies for targets: http://make.paulandlesley.org/autodep.html
33 * \todo Investigate whether switching to cmake as cross-platform build system would be a good idea.
34 * \todo Switch from nmake to GNU make under Windows http://gnuwin32.sourceforge.net/
35 *
36 * \todo Replace VarSets by SmallSet<size_t> where appropriate, in order to minimize the use of FactorGraph::findVar().
37 *
38 * \idea Disentangle structures. In particular, ensure that graphical properties are not
39 * entangled with probabilistic properties. For example, a FactorGraph contains several
40 * components:
41 * - a BipartiteGraph
42 * - an array of variable labels
43 * - an array of variable state space sizes
44 * - an array of pointers to factor value vectors
45 * In this way, each factor could be implemented differently, e.g., we could have
46 * some sparse factors, some noisy-OR factors, some dense factors, some arbitrary
47 * precision factors, etc.
48 *
49 * \idea Use Boost::uBLAS framework to deal with matrices, especially, with 2D sparse matrices.
50 * See http://www.boost.org/libs/numeric/ublas/doc/matrix_sparse.htm
51 *
52 * \idea Introduce naming scheme:
53 * - all Vars should be named v_..., e.g. v_i instead of i
54 * - all VarSets should be named vs_..., e.g. v_i instead of i
55 * - all Factors should be named f_..., e.g. f_I instead of I
56 * - all indices should be named _..., e.g. _k instead of k
57 * - all iterators should be named i_, e.g. i_i is an iterator to i
58 * - all const_iterators should be named ci_, e.g. ci_i is an iterator to i
59 **/
60
61
62 /** \page discussion Discussion of possible improvements
63 * \section discuss_extendedgraphs Extended factorgraphs/regiongraphs
64 *
65 * A FactorGraph and a RegionGraph are often equipped with
66 * additional properties for nodes and edges. The code to initialize those
67 * is often quite similar. Maybe one could abstract this, e.g.:
68 * \code
69 * template <typename Node1Properties, typename Node2Properties, typename EdgeProperties>
70 * class ExtFactorGraph : public FactorGraph {
71 * public:
72 * std::vector<Node1Properties> node1Props;
73 * std::vector<Node2Properties> node2Props;
74 * std::vector<std::vector<EdgeProperties> > edgeProps;
75 * // ...
76 * }
77 * \endcode
78 *
79 * Advantages:
80 * - Less code duplication.
81 * - Easier maintainability.
82 * - Easier to write new inference algorithms.
83 *
84 * Disadvantages:
85 * - Cachability may be worse.
86 * - A problem is the case where there are no properties for either type of nodes or for edges.
87 * Maybe this can be solved using specializations, or using variadac template arguments?
88 * Another possible solution would be to define a "class Empty {}", and add some code
89 * that checks for the typeid, comparing it with Empty, and doing something special in that case
90 * (e.g., not allocating memory).
91 * - The main disadvantage of this approach seems to be that it leads to even more entanglement.
92 * Therefore this is probably a bad idea.
93 *
94 * \section discuss_templates Polymorphism by template parameterization
95 * Instead of polymorphism by inheritance, use polymorphism by template parameterization.
96 * For example, the real reason for introducing the complicated inheritance scheme of InfAlg
97 * was for functions like dai::calcMarginal. Instead, one could use a template function:
98 * \code
99 * template<typename InfAlg>
100 * Factor calcMarginal( const InfAlg &obj, const VarSet &ns, bool reInit );
101 * \endcode
102 * This would assume that the type InfAlg supports certain methods. Ideally, one would use
103 * concepts to define different classes of inference algorithms with different capabilities,
104 * for example the ability to calculate logZ, the ability to calculate marginals, the ability to
105 * calculate bounds, the ability to calculate MAP states, etc. Then, one would use traits
106 * classes in order to be able to query the capabilities of the model. For example, one would be
107 * able to query whether the inference algorithm supports calculation of logZ. Unfortunately,
108 * this is compile-time polymorphism, whereas tests/testdai needs runtime polymorphism.
109 * Therefore this is probably a bad idea.
110 */
111
112
113 /** \mainpage libDAI reference manual
114 * \author Joris Mooij
115 * \version git HEAD
116 * \date October 10, 2008
117 *
118 * \section about About libDAI
119 * libDAI is a free/open source C++ library (licensed under GPL) that provides
120 * implementations of various (approximate) inference methods for discrete
121 * graphical models. libDAI supports arbitrary factor graphs with discrete
122 * variables; this includes discrete Markov Random Fields and Bayesian
123 * Networks.
124 *
125 * The library is targeted at researchers; to be able to use the library, a
126 * good understanding of graphical models is needed.
127 *
128 * \section limitations Limitations
129 * libDAI is not intended to be a complete package for approximate inference.
130 * Instead, it should be considered as an "inference engine", providing
131 * various inference methods. In particular, it contains no GUI, currently
132 * only supports its own file format for input and output (although support
133 * for standard file formats may be added later), and provides very limited
134 * visualization functionalities.
135 *
136 * \section features Features
137 * Currently, libDAI supports the following (approximate) inference methods:
138 * - Exact inference by brute force enumeration;
139 * - Exact inference by junction-tree methods;
140 * - Mean Field;
141 * - Loopy Belief Propagation [\ref KFL01];
142 * - Tree Expectation Propagation [\ref MiQ04];
143 * - Generalized Belief Propagation [\ref YFW05];
144 * - Double-loop GBP [\ref HAK03];
145 * - Various variants of Loop Corrected Belief Propagation
146 * [\ref MoK07, \ref MoR05];
147 * - Gibbs sampler.
148 *
149 * \section language Why C++?
150 * Because libDAI is implemented in C++, it is very fast compared with
151 * implementations in MatLab (a factor 1000 faster is not uncommon).
152 * libDAI does provide a MatLab interface for easy integration with MatLab.
153 */
154
155
156 /** \example example.cpp
157 */
158
159
160 /** \page quickstart Quick start
161 * An example program illustrating basic usage of libDAI is given in examples/example.cpp.
162 */
163
164
165 /** \page biboliography Bibliography
166 * \section Bibliograpy
167 * \anchor KFL01 \ref KFL01
168 * F. R. Kschischang and B. J. Frey and H.-A. Loeliger (2001):
169 * "Factor Graphs and the Sum-Product Algorithm",
170 * <em>IEEE Transactions on Information Theory</em> 47(2):498-519.
171 * http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=910572
172 *
173 * \anchor MiQ04 \ref MiQ04
174 * T. Minka and Y. Qi (2004):
175 * "Tree-structured Approximations by Expectation Propagation",
176 * <em>Advances in Neural Information Processing Systems</em> (NIPS) 16.
177 * http://books.nips.cc/papers/files/nips16/NIPS2003_AA25.pdf
178 *
179 * \anchor MoR05 \ref MoR05
180 * A. Montanari and T. Rizzo (2005):
181 * "How to Compute Loop Corrections to the Bethe Approximation",
182 * <em>Journal of Statistical Mechanics: Theory and Experiment</em>
183 * 2005(10)-P10011.
184 * http://stacks.iop.org/1742-5468/2005/P10011
185 *
186 * \anchor YFW05 \ref YFW05
187 * J. S. Yedidia and W. T. Freeman and Y. Weiss (2005):
188 * "Constructing Free-Energy Approximations and Generalized Belief Propagation Algorithms",
189 * <em>IEEE Transactions on Information Theory</em>
190 * 51(7):2282-2312.
191 * http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1459044
192 *
193 * \anchor HAK03 \ref HAK03
194 * T. Heskes and C. A. Albers and H. J. Kappen (2003):
195 * "Approximate Inference and Constrained Optimization",
196 * <em>Proceedings of the 19th Annual Conference on Uncertainty in Artificial Intelligence (UAI-03)</em> pp. 313-320.
197 * http://www.snn.ru.nl/reports/Heskes.uai2003.ps.gz
198 *
199 * \anchor MoK07 \ref MoK07
200 * J. M. Mooij and H. J. Kappen (2007):
201 * "Loop Corrections for Approximate Inference on Factor Graphs",
202 * <em>Journal of Machine Learning Research</em> 8:1113-1143.
203 * http://www.jmlr.org/papers/volume8/mooij07a/mooij07a.pdf
204 *
205 * \anchor MoK07b \ref MoK07b
206 * J. M. Mooij and H. J. Kappen (2007):
207 * "Sufficient Conditions for Convergence of the Sum-Product Algorithm",
208 * <em>IEEE Transactions on Information Theory</em> 53(12):4422-4437.
209 * http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=4385778
210 *
211 * \anchor EaG09 \ref EaG09
212 * F. Eaton and Z. Ghahramani (2009):
213 * "Choosing a Variable to Clamp",
214 * <em>Proceedings of the Twelfth International Conference on Artificial Intelligence and Statistics (AISTATS 2009)</em> 5:145-152
215 * http://jmlr.csail.mit.edu/proceedings/papers/v5/eaton09a/eaton09a.pdf
216 */
217
218
219 /** \page fileformat libDAI factorgraph file format
220 *
221 * This page describes the .fg fileformat used in libDAI to store factor graphs.
222 * Markov Random Fields are special cases of factor graphs, as are Bayesian
223 * networks. A factor graph can be specified as follows: for each factor, one has
224 * to specify which variables occur in the factor, what their respective
225 * cardinalities (i.e., number of possible values) are, and a table listing all
226 * the values of that factor for all possible configurations of these variables.
227 * A .fg file is not much more than that. It starts with a line containing the
228 * number of factors in that graph, followed by an empty line. Then all factors
229 * are specified, one block for each factor, where the blocks are seperated by
230 * empty lines. Each variable occurring in the factor graph has a unique
231 * identifier, its index (which should be a nonnegative integer). Comment lines
232 * start with #.
233 *
234 * Each block starts with a line containing the number of variables in that
235 * factor. The second line contains the indices of these variables, seperated by
236 * spaces (indices are nonnegative integers and to avoid confusion, it is
237 * suggested to start counting at 0). The third line contains the number of
238 * possible values of each of these variables, also seperated by spaces. Note that
239 * there is some redundancy here, since if a variable appears in more than one
240 * factor, the cardinality of that variable appears several times in the .fg file.
241 * The fourth line contains the number of nonzero entries in the factor table.
242 * The rest of the lines contain these nonzero entries; each entry consists of a
243 * table index, followed by white-space, followed by the value corresponding to
244 * that table index. The most difficult part is getting the indexing right. The
245 * convention that is used is that the left-most variables cycle through their
246 * values the fastest (similar to MATLAB indexing of multidimensional arrays). An
247 * example block describing one factor is:
248 *
249 * 3\n
250 * 4 8 7\n
251 * 3 2 2\n
252 * 11\n
253 * 0 0.1\n
254 * 1 3.5\n
255 * 2 2.8\n
256 * 3 6.3\n
257 * 4 8.4\n
258 * 6 7.4\n
259 * 7 2.4\n
260 * 8 8.9\n
261 * 9 1.3\n
262 * 10 1.6\n
263 * 12 6.4\n
264 * 11 2.6\n
265 *
266 * which corresponds to the following factor:
267 *
268 * \f[
269 * \begin{array}{ccc|c}
270 * x_4 & x_8 & x_7 & \mbox{value}\\
271 * \hline
272 * 0 & 0 & 0 & 0.1\\
273 * 1 & 0 & 0 & 3.5\\
274 * 2 & 0 & 0 & 2.8\\
275 * 0 & 1 & 0 & 6.3\\
276 * 1 & 1 & 0 & 8.4\\
277 * 2 & 1 & 0 & 0.0\\
278 * 0 & 0 & 1 & 7.4\\
279 * 1 & 0 & 1 & 2.4\\
280 * 2 & 0 & 1 & 8.9\\
281 * 0 & 1 & 1 & 1.3\\
282 * 1 & 1 & 1 & 1.6\\
283 * 2 & 1 & 1 & 2.6
284 * \end{array}
285 * \f]
286 *
287 * Note that the value of x_4 changes fastest, followed by that of x_8, and x_7
288 * varies the slowest, corresponding to the second line of the block ("4 8 7").
289 * Further, x_4 can take on three values, and x_8 and x_7 each have two possible
290 * values, as described in the third line of the block ("3 2 2"). The table
291 * contains 11 non-zero entries (all except for the fifth entry). Note that the
292 * eleventh and twelveth entries are interchanged.
293 *
294 * A final note: the internal representation in libDAI of the factor above is
295 * different, because the variables are ordered according to their indices
296 * (i.e., the ordering would be x_4 x_7 x_8) and the values of the table are
297 * stored accordingly, with the variable having the smallest index changing
298 * fastest:
299 *
300 * \f[
301 * \begin{array}{ccc|c}
302 * x_4 & x_7 & x_8 & \mbox{value}\\
303 * \hline
304 * 0 & 0 & 0 & 0.1\\
305 * 1 & 0 & 0 & 3.5\\
306 * 2 & 0 & 0 & 2.8\\
307 * 0 & 1 & 0 & 7.4\\
308 * 1 & 1 & 0 & 2.4\\
309 * 2 & 1 & 0 & 8.9\\
310 * 0 & 0 & 1 & 6.3\\
311 * 1 & 0 & 1 & 8.4\\
312 * 2 & 0 & 1 & 0.0\\
313 * 0 & 1 & 1 & 1.3\\
314 * 1 & 1 & 1 & 1.6\\
315 * 2 & 1 & 1 & 2.6
316 * \end{array}
317 * \f]
318 */
319
320
321 /** \page license License
322 * \verbinclude COPYING
323 */