Merge branch 'joris'
[libdai.git] / include / dai / daialg.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 Defines abstract base class InfAlg, its descendants DAIAlg<T>, the specializations DAIAlgFG and DAIAlgRG and some generic inference methods.
25 /// \todo Improve documentation
26
27
28 #ifndef __defined_libdai_daialg_h
29 #define __defined_libdai_daialg_h
30
31
32 #include <string>
33 #include <iostream>
34 #include <vector>
35 #include <dai/factorgraph.h>
36 #include <dai/regiongraph.h>
37
38
39 namespace dai {
40
41
42 /// InfAlg is an abstract base class, defining the common interface of all inference algorithms in libDAI.
43 /** \todo General marginalization functions like calcMarginal now copy a complete InfAlg object. Instead,
44 * it would make more sense that they construct a new object without copying the FactorGraph or RegionGraph.
45 * Or they can simply be made methods of the general InfAlg class.
46 * \idea Use a PropertySet as output of an InfAlg, instead of functions like maxDiff() and Iterations().
47 */
48 class InfAlg {
49 public:
50 /// Virtual desctructor (needed because this class contains virtual functions)
51 virtual ~InfAlg() {}
52
53 public:
54 /// Returns a pointer to a new, cloned copy of *this (i.e., virtual copy constructor)
55 virtual InfAlg* clone() const = 0;
56
57 /// Returns a pointer to a newly constructed object *this (i.e., virtual default constructor)
58 virtual InfAlg* create() const = 0;
59
60 /// Identifies itself for logging purposes
61 virtual std::string identify() const = 0;
62
63 /// Returns the "belief" (i.e., approximate marginal probability distribution) of a variable
64 virtual Factor belief( const Var &n ) const = 0;
65
66 /// Returns the "belief" (i.e., approximate marginal probability distribution) of a set of variables
67 virtual Factor belief( const VarSet &n ) const = 0;
68
69 /// Returns all "beliefs" (i.e., approximate marginal probability distribution) calculated by the algorithm
70 virtual std::vector<Factor> beliefs() const = 0;
71
72 /// Returns the logarithm of the (approximated) partition sum (normalizing constant of the factor graph)
73 virtual Real logZ() const = 0;
74
75 /// Initializes all data structures of the approximate inference algorithm
76 /** This method should be called at least once before run() is called
77 */
78 virtual void init() = 0;
79
80 /// Initializes all data structures corresponding to some set of variables
81 /** This method can be used to do a partial initialization after a part of the factor graph has changed.
82 * Instead of initializing all data structures, it only initializes those involving the variables in ns.
83 */
84 virtual void init( const VarSet &ns ) = 0;
85
86 /// Runs the approximate inference algorithm
87 /* Before run() is called the first time, init() should be called.
88 * If run() returns successfully, the results can be queried using the methods belief(), beliefs() and logZ().
89 */
90 virtual double run() = 0;
91
92 /// Clamp variable n to value i (i.e. multiply with a Kronecker delta \f$\delta_{x_n, i}\f$)
93 virtual void clamp( const Var & n, size_t i, bool backup = false ) = 0;
94
95 /// Set all factors interacting with var(i) to 1
96 virtual void makeCavity( size_t i, bool backup = false ) = 0;
97
98 /// Return maximum difference between single node beliefs in the last pass
99 /// \throw Exception if not implemented/supported
100 virtual double maxDiff() const = 0;
101
102 /// Return number of passes over the factorgraph
103 /// \throw Exception if not implemented/supported
104 virtual size_t Iterations() const = 0;
105
106
107 /// Get reference to underlying FactorGraph
108 virtual FactorGraph &fg() = 0;
109
110 /// Get const reference to underlying FactorGraph
111 virtual const FactorGraph &fg() const = 0;
112
113 /// Save factor I
114 virtual void backupFactor( size_t I ) = 0;
115 /// Save Factors involving ns
116 virtual void backupFactors( const VarSet &ns ) = 0;
117
118 /// Restore factor I
119 virtual void restoreFactor( size_t I ) = 0;
120 /// Restore Factors involving ns
121 virtual void restoreFactors( const VarSet &ns ) = 0;
122 };
123
124
125 /// Combines an InfAlg and a graphical model, e.g., a FactorGraph or RegionGraph
126 /** \tparam GRM Should be castable to FactorGraph
127 * \todo A DAIAlg should not inherit from a FactorGraph or RegionGraph, but should
128 * store a reference to the graphical model object. This prevents needless copying
129 * of (possibly large) data structures. Disadvantage: the caller must not change
130 * the graphical model between calls to the inference algorithm (maybe a smart_ptr
131 * or some locking mechanism would help here?).
132 */
133 template <class GRM>
134 class DAIAlg : public InfAlg, public GRM {
135 public:
136 /// Default constructor
137 DAIAlg() : InfAlg(), GRM() {}
138
139 /// Construct from GRM
140 DAIAlg( const GRM &grm ) : InfAlg(), GRM(grm) {}
141
142 /// Copy constructor
143 DAIAlg( const DAIAlg & x ) : InfAlg(x), GRM(x) {}
144
145 /// Assignment operator
146 DAIAlg & operator=( const DAIAlg &x ) {
147 if( this != &x ) {
148 InfAlg::operator=(x);
149 GRM::operator=(x);
150 }
151 return *this;
152 }
153
154 /// Save factor I
155 void backupFactor( size_t I ) { GRM::backupFactor( I ); }
156 /// Save Factors involving ns
157 void backupFactors( const VarSet &ns ) { GRM::backupFactors( ns ); }
158
159 /// Restore factor I
160 void restoreFactor( size_t I ) { GRM::restoreFactor( I ); }
161 /// Restore Factors involving ns
162 void restoreFactors( const VarSet &ns ) { GRM::restoreFactors( ns ); }
163
164 /// Clamp variable n to value i (i.e. multiply with a Kronecker delta \f$\delta_{x_n, i}\f$)
165 void clamp( const Var & n, size_t i, bool backup = false ) { GRM::clamp( n, i, backup ); }
166
167 /// Set all factors interacting with var(i) to 1
168 void makeCavity( size_t i, bool backup = false ) { GRM::makeCavity( i, backup ); }
169
170 /// Get reference to underlying FactorGraph
171 FactorGraph &fg() { return (FactorGraph &)(*this); }
172
173 /// Get const reference to underlying FactorGraph
174 const FactorGraph &fg() const { return (const FactorGraph &)(*this); }
175 };
176
177
178 /// Base class for inference algorithms that operate on a FactorGraph
179 typedef DAIAlg<FactorGraph> DAIAlgFG;
180
181 /// Base class for inference algorithms that operate on a RegionGraph
182 typedef DAIAlg<RegionGraph> DAIAlgRG;
183
184
185 Factor calcMarginal( const InfAlg & obj, const VarSet & ns, bool reInit );
186 std::vector<Factor> calcPairBeliefs( const InfAlg & obj, const VarSet& ns, bool reInit );
187 std::vector<Factor> calcPairBeliefsNew( const InfAlg & obj, const VarSet& ns, bool reInit );
188 Factor calcMarginal2ndO( const InfAlg & obj, const VarSet& ns, bool reInit );
189
190
191 } // end of namespace dai
192
193
194 #endif