Binary Bayes Net, incomplete

cpp/BinaryConditional.h

@@ -0,0 +1,88 @@
+/**
+ * @file    DiscreteConditional.h
+ * @brief   Discrete Conditional node for use in Bayes nets
+ * @author  Manohar Paluri
+ */
+
+// \callgraph
+
+#pragma once
+
+#include <list>
+#include <string>
+#include <iostream>
+#include <boost/shared_ptr.hpp>
+#include <boost/foreach.hpp> // TODO: make cpp file
+#include <boost/serialization/list.hpp>
+#include "Conditional.h"
+
+namespace gtsam {
+
+	/**
+	 * Conditional node for use in a Bayes net
+	 */
+	class BinaryConditional: public Conditional {
+
+	private:
+
+		std::list<std::string> parents_;
+
+	public:
+
+		/** convenience typename for a shared pointer to this class */
+		typedef boost::shared_ptr<BinaryConditional> shared_ptr;
+
+		/**
+		 * Empty Constructor to make serialization possible
+		 */
+		BinaryConditional(){}
+
+		/**
+		 * No parents
+		 */
+		BinaryConditional(const std::string& key, double p) :
+			Conditional(key) {
+		}
+
+		/**
+		 * Single parent
+		 */
+		BinaryConditional(const std::string& key, const std::string& parent, const std::vector<double>& cpt) :
+			Conditional(key) {
+			parents_.push_back(parent);
+		}
+
+		/** print */
+		void print(const std::string& s = "BinaryConditional") const {
+			std::cout << s << " P(" << key_;
+			if (parents_.size()>0) std::cout << " |";
+			BOOST_FOREACH(std::string parent, parents_) std::cout << " " << parent;
+			std::cout << ")" << std::endl;
+		}
+
+		/** check equality */
+		bool equals(const Conditional& c, double tol = 1e-9) const {
+			if (!Conditional::equals(c)) return false;
+			const BinaryConditional* p = dynamic_cast<const BinaryConditional*> (&c);
+			if (p == NULL) return false;
+			return parents_ == p->parents_;
+		}
+
+		/** return parents */
+		std::list<std::string> parents() const { return parents_;}
+
+		/** find the number of parents */
+		size_t nrParents() const {
+			return parents_.size();
+		}
+
+	private:
+		/** Serialization function */
+		friend class boost::serialization::access;
+		template<class Archive>
+		void serialize(Archive & ar, const unsigned int version) {
+			ar & boost::serialization::make_nvp("Conditional", boost::serialization::base_object<Conditional>(*this));
+			ar & BOOST_SERIALIZATION_NVP(parents_);
+		}
+	};
+} /// namespace gtsam

cpp/Makefile.am

@@ -85,6 +85,12 @@ testSymbolicFactorGraph_LDADD   = libgtsam.la
 testSymbolicBayesNet_SOURCES    = $(example) testSymbolicBayesNet.cpp
 testSymbolicBayesNet_LDADD      = libgtsam.la
 
+# Binary Inference
+headers += BinaryConditional.h
+check_PROGRAMS += testBinaryBayesNet
+testBinaryBayesNet_SOURCES    = testBinaryBayesNet.cpp
+testBinaryBayesNet_LDADD      = libgtsam.la 
+
 # Gaussian inference 
 headers += GaussianFactorSet.h
 sources += VectorConfig.cpp GaussianFactor.cpp GaussianFactorGraph.cpp GaussianConditional.cpp GaussianBayesNet.cpp

cpp/testBinaryBayesNet.cpp

@@ -1,27 +1,72 @@
 /**
  * @file    testBinaryBayesNet.cpp
- * @brief   Unit tests for Bayes Tree
- * @author  Frank Dellaert
+ * @brief   Unit tests for BinaryBayesNet
+ * @author  Manohar Paluri
  */
 
+// STL/C++
+#include <iostream>
+#include <sstream>
 #include <CppUnitLite/TestHarness.h>
+#include <boost/tuple/tuple.hpp>
+#include <boost/foreach.hpp>
 
+#include <boost/assign/std/vector.hpp> // for operator +=
+using namespace boost::assign;
+
+#ifdef HAVE_BOOST_SERIALIZATION
+#include <boost/archive/text_oarchive.hpp>
+#include <boost/archive/text_iarchive.hpp>
+#endif //HAVE_BOOST_SERIALIZATION
+
+#include "BinaryConditional.h"
+#include "BayesNet-inl.h"
+#include "smallExample.h"
+#include "Ordering.h"
+
+using namespace std;
 using namespace gtsam;
 
+/** A Bayes net made from binary conditional probability tables */
+typedef BayesNet<BinaryConditional> BinaryBayesNet;
+
+struct BinaryConfig {
+	bool px_;
+	bool py_;
+
+	BinaryConfig( bool px, bool py ):px_(px), py_(py){}
+};
+
+double probability(const BinaryBayesNet& bayesNet, const BinaryConfig& config) {
+
+	return 0;
+}
+
 /* ************************************************************************* */
 TEST( BinaryBayesNet, constructor )
 {
-	map<string,BinaryCPT> tables;
-	BinaryCPT pA(0.01);tables.insert("A",pA);
-	BinaryCPT pB("S",0.6,0.3);
-	BinaryBayesNet binaryBayesNet(tables);
-	BinaryConfig allFalse(false,false,false,...);
-	DOUBLES_EQUAL(0.12,binaryBayesNet.probability(allFalse));
+  // small Bayes Net x <- y
+  // p(y) = 0.2
+  // p(x|y=0) = 0.3
+  // p(x|y=1) = 0.5
+
+  // unary conditional for y
+  boost::shared_ptr<BinaryConditional> py(new BinaryConditional("y",0.2));
+
+  // single parent conditional for x
+  vector<double> cpt;
+  cpt += 0.3, 0.5; // array index corresponds to binary parent configuration
+  boost::shared_ptr<BinaryConditional> px_y(new BinaryConditional("x","y",cpt));
+
+  // push back conditionals in topological sort order (parents last)
+  BinaryBayesNet bbn;
+  bbn.push_back(py);
+  bbn.push_back(px_y);
+
+  // Test probability of 00,01,10,11
+  //DOUBLES_EQUAL(0.56,probability(bbn,BinaryConfig(false,false)),0.01); // P(y=0)P(x=0|y=0) = 0.8 * 0.7 = 0.56;
 }
 
 /* ************************************************************************* */
-int main() {
-	TestResult tr;
-	return TestRegistry::runAllTests(tr);
-}
+int main() { TestResult tr; return TestRegistry::runAllTests(tr);}
 /* ************************************************************************* */