Fixed tests to work with new definition of error.

gtsam/hybrid/tests/testGaussianMixture.cpp

@@ -118,9 +118,10 @@ TEST(GaussianMixture, Error) {
   values.insert(X(2), Vector2::Zero());
   auto error_tree = mixture.error(values);
 
-  // regression
+  // Check result.
   std::vector<DiscreteKey> discrete_keys = {m1};
-  std::vector<double> leaves = {0.5, 4.3252595};
+  std::vector<double> leaves = {conditional0->error(values),
+                                conditional1->error(values)};
   AlgebraicDecisionTree<Key> expected_error(discrete_keys, leaves);
 
   EXPECT(assert_equal(expected_error, error_tree, 1e-6));
@@ -128,10 +129,11 @@ TEST(GaussianMixture, Error) {
   // Regression for non-tree version.
   DiscreteValues assignment;
   assignment[M(1)] = 0;
-  EXPECT_DOUBLES_EQUAL(0.5, mixture.error({values, assignment}), 1e-8);
+  EXPECT_DOUBLES_EQUAL(conditional0->error(values),
+                       mixture.error({values, assignment}), 1e-8);
   assignment[M(1)] = 1;
-  EXPECT_DOUBLES_EQUAL(4.3252595155709335, mixture.error({values, assignment}),
-                       1e-8);
+  EXPECT_DOUBLES_EQUAL(conditional1->error(values),
+                       mixture.error({values, assignment}), 1e-8);
 }
 
 /* ************************************************************************* */

gtsam/hybrid/tests/testHybridBayesNet.cpp

@@ -217,23 +217,22 @@ TEST(HybridBayesNet, Error) {
   auto error_tree = hybridBayesNet->error(delta.continuous());
 
   std::vector<DiscreteKey> discrete_keys = {{M(0), 2}, {M(1), 2}};
-  std::vector<double> leaves = {0.0097568009, 3.3973404e-31, 0.029126214,
-                                0.0097568009};
+  std::vector<double> leaves = {-4.1609374, -4.1706942, -4.141568, -4.1609374};
   AlgebraicDecisionTree<Key> expected_error(discrete_keys, leaves);
 
   // regression
-  EXPECT(assert_equal(expected_error, error_tree, 1e-9));
+  EXPECT(assert_equal(expected_error, error_tree, 1e-6));
 
   // Error on pruned Bayes net
   auto prunedBayesNet = hybridBayesNet->prune(2);
   auto pruned_error_tree = prunedBayesNet.error(delta.continuous());
 
-  std::vector<double> pruned_leaves = {2e50, 3.3973404e-31, 2e50, 0.0097568009};
+  std::vector<double> pruned_leaves = {2e50, -4.1706942, 2e50, -4.1609374};
   AlgebraicDecisionTree<Key> expected_pruned_error(discrete_keys,
                                                    pruned_leaves);
 
   // regression
-  EXPECT(assert_equal(expected_pruned_error, pruned_error_tree, 1e-9));
+  EXPECT(assert_equal(expected_pruned_error, pruned_error_tree, 1e-6));
 
   // Verify error computation and check for specific error value
   DiscreteValues discrete_values{{M(0), 1}, {M(1), 1}};

gtsam/linear/tests/testGaussianConditional.cpp

@@ -381,13 +381,13 @@ TEST(GaussianConditional, FromMeanAndStddev) {
   auto conditional1 =
       GaussianConditional::FromMeanAndStddev(X(0), A1, X(1), b, sigma);
   Vector2 e1 = (x0 - (A1 * x1 + b)) / sigma;
-  double expected1 = 0.5 * e1.dot(e1);
+  double expected1 = 0.5 * e1.dot(e1) - conditional1.logNormalizationConstant();
   EXPECT_DOUBLES_EQUAL(expected1, conditional1.error(values), 1e-9);
 
   auto conditional2 = GaussianConditional::FromMeanAndStddev(X(0), A1, X(1), A2,
                                                              X(2), b, sigma);
   Vector2 e2 = (x0 - (A1 * x1 + A2 * x2 + b)) / sigma;
-  double expected2 = 0.5 * e2.dot(e2);
+  double expected2 = 0.5 * e2.dot(e2) - conditional2.logNormalizationConstant();
   EXPECT_DOUBLES_EQUAL(expected2, conditional2.error(values), 1e-9);
 }
 
@@ -448,20 +448,23 @@ TEST(GaussianConditional, sample) {
 }
 
 /* ************************************************************************* */
-TEST(GaussianConditional, LogNormalizationConstant) {
+TEST(GaussianConditional, Error) {
   // Create univariate standard gaussian conditional
-  auto std_gaussian =
+  auto stdGaussian =
       GaussianConditional::FromMeanAndStddev(X(0), Vector1::Zero(), 1.0);
   VectorValues values;
   values.insert(X(0), Vector1::Zero());
-  double logDensity = std_gaussian.logDensity(values);
+  double error = stdGaussian.error(values);
 
   // Regression.
   // These values were computed by hand for a univariate standard gaussian.
-  EXPECT_DOUBLES_EQUAL(-0.9189385332046727, logDensity, 1e-9);
-  EXPECT_DOUBLES_EQUAL(0.3989422804014327, exp(logDensity), 1e-9);
+  EXPECT_DOUBLES_EQUAL(0.9189385332046727, error, 1e-9);
+  EXPECT_DOUBLES_EQUAL(0.3989422804014327, exp(-error), 1e-9);
+}
 
-  // Similar test for multivariate gaussian but with sigma 2.0
+/* ************************************************************************* */
+// Similar test for multivariate gaussian but with sigma 2.0
+TEST(GaussianConditional, LogNormalizationConstant) {
   double sigma = 2.0;
   auto conditional = GaussianConditional::FromMeanAndStddev(X(0), Vector3::Zero(), sigma);
   VectorValues x;
@@ -469,7 +472,8 @@ TEST(GaussianConditional, LogNormalizationConstant) {
   Matrix3 Sigma = I_3x3 * sigma * sigma;
   double expectedLogNormalizingConstant = log(1 / sqrt((2 * M_PI * Sigma).determinant()));
 
-  EXPECT_DOUBLES_EQUAL(expectedLogNormalizingConstant, conditional.logNormalizationConstant(), 1e-9);
+  EXPECT_DOUBLES_EQUAL(expectedLogNormalizingConstant,
+                       conditional.logNormalizationConstant(), 1e-9);
 }
 
 /* ************************************************************************* */

gtsam/linear/tests/testGaussianDensity.cpp

@@ -52,7 +52,7 @@ TEST(GaussianDensity, FromMeanAndStddev) {
 
   auto density = GaussianDensity::FromMeanAndStddev(X(0), b, sigma);
   Vector2 e = (x0 - b) / sigma;
-  double expected = 0.5 * e.dot(e);
+  double expected = 0.5 * e.dot(e) - density.logNormalizationConstant();
   EXPECT_DOUBLES_EQUAL(expected, density.error(values), 1e-9);
 }