Added the example graph in powerMethodExample.h

gtsam/linear/tests/powerMethodExample.h

@@ -0,0 +1,67 @@
+/* ----------------------------------------------------------------------------
+
+ * GTSAM Copyright 2010-2019, Georgia Tech Research Corporation,
+ * Atlanta, Georgia 30332-0415
+ * All Rights Reserved
+ * Authors: Frank Dellaert, et al. (see THANKS for the full author list)
+
+ * See LICENSE for the license information
+
+ * -------------------------------------------------------------------------- */
+
+/**
+ * powerMethodExample.h
+ *
+ * @file   powerMethodExample.h
+ * @date   Nov 2020
+ * @author Jing Wu
+ * @brief  Create sparse and dense factor graph for
+ * PowerMethod/AcceleratedPowerMethod
+ */
+
+#include <gtsam/inference/Symbol.h>
+
+#include <iostream>
+
+
+namespace gtsam {
+namespace linear {
+namespace test {
+namespace example {
+
+/* ************************************************************************* */
+inline GaussianFactorGraph createSparseGraph() {
+  using symbol_shorthand::X;
+  // Let's make a scalar synchronization graph with 4 nodes
+  GaussianFactorGraph fg;
+  auto model = noiseModel::Unit::Create(1);
+  for (size_t j = 0; j < 3; j++) {
+    fg.add(X(j), -I_1x1, X(j + 1), I_1x1, Vector1::Zero(), model);
+  }
+  fg.add(X(3), -I_1x1, X(0), I_1x1, Vector1::Zero(), model);  // extra row
+
+  return fg;
+}
+
+/* ************************************************************************* */
+inline GaussianFactorGraph createDenseGraph() {
+  using symbol_shorthand::X;
+  // Let's make a scalar synchronization graph with 10 nodes
+  GaussianFactorGraph fg;
+  auto model = noiseModel::Unit::Create(1);
+  // Iterate over nodes
+  for (size_t j = 0; j < 10; j++) {
+    // Each node has an edge with all the others
+    for (size_t i = 1; i < 10; i++)
+    fg.add(X(j), -I_1x1, X((j + i) % 10), I_1x1, Vector1::Zero(), model);
+  }
+
+  return fg;
+}
+
+/* ************************************************************************* */
+
+}  // namespace example
+}  // namespace test
+}  // namespace linear
+}  // namespace gtsam

gtsam/linear/tests/testAcceleratedPowerMethod.cpp

@@ -24,6 +24,7 @@
 #include <gtsam/inference/Symbol.h>
 #include <gtsam/linear/AcceleratedPowerMethod.h>
 #include <gtsam/linear/GaussianFactorGraph.h>
+#include <gtsam/linear/tests/powerMethodExample.h>
 
 #include <Eigen/Core>
 #include <Eigen/Dense>
@@ -33,7 +34,6 @@
 
 using namespace std;
 using namespace gtsam;
-using symbol_shorthand::X;
 
 /* ************************************************************************* */
 TEST(AcceleratedPowerMethod, acceleratedPowerIteration) {
@@ -65,12 +65,7 @@ TEST(AcceleratedPowerMethod, acceleratedPowerIteration) {
 /* ************************************************************************* */
 TEST(AcceleratedPowerMethod, useFactorGraphSparse) {
   // Let's make a scalar synchronization graph with 4 nodes
-  GaussianFactorGraph fg;
+  GaussianFactorGraph fg = gtsam::linear::test::example::createSparseGraph();
-  auto model = noiseModel::Unit::Create(1);
-  for (size_t j = 0; j < 3; j++) {
-    fg.add(X(j), -I_1x1, X(j + 1), I_1x1, Vector1::Zero(), model);
-  }
-  fg.add(X(3), -I_1x1, X(0), I_1x1, Vector1::Zero(), model);  // extra row
 
   // Get eigenvalues and eigenvectors with Eigen
   auto L = fg.hessian();
@@ -105,20 +100,7 @@ TEST(AcceleratedPowerMethod, useFactorGraphSparse) {
 /* ************************************************************************* */
 TEST(AcceleratedPowerMethod, useFactorGraphDense) {
   // Let's make a scalar synchronization graph with 10 nodes
-  GaussianFactorGraph fg;
+  GaussianFactorGraph fg = gtsam::linear::test::example::createDenseGraph();
-  auto model = noiseModel::Unit::Create(1);
-  // Each node has an edge with all the others
-  for (size_t j = 0; j < 10; j++) {
-    fg.add(X(j), -I_1x1, X((j + 1)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 2)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 3)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 4)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 5)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 6)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 7)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 8)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 9)%10 ), I_1x1, Vector1::Zero(), model);
-  }
 
   // Get eigenvalues and eigenvectors with Eigen
   auto L = fg.hessian();

gtsam/linear/tests/testPowerMethod.cpp

@@ -24,6 +24,7 @@
 #include <gtsam/inference/Symbol.h>
 #include <gtsam/linear/GaussianFactorGraph.h>
 #include <gtsam/linear/PowerMethod.h>
+#include <gtsam/linear/tests/powerMethodExample.h>
 
 #include <Eigen/Core>
 #include <Eigen/Dense>
@@ -33,7 +34,6 @@
 
 using namespace std;
 using namespace gtsam;
-using symbol_shorthand::X;
 
 /* ************************************************************************* */
 TEST(PowerMethod, powerIteration) {
@@ -63,12 +63,7 @@ TEST(PowerMethod, powerIteration) {
 /* ************************************************************************* */
 TEST(PowerMethod, useFactorGraphSparse) {
   // Let's make a scalar synchronization graph with 4 nodes
-  GaussianFactorGraph fg;
+  GaussianFactorGraph fg = gtsam::linear::test::example::createSparseGraph();
-  auto model = noiseModel::Unit::Create(1);
-  for (size_t j = 0; j < 3; j++) {
-    fg.add(X(j), -I_1x1, X(j + 1), I_1x1, Vector1::Zero(), model);
-  }
-  fg.add(X(3), -I_1x1, X(0), I_1x1, Vector1::Zero(), model);  // extra row
 
   // Get eigenvalues and eigenvectors with Eigen
   auto L = fg.hessian();
@@ -96,20 +91,7 @@ TEST(PowerMethod, useFactorGraphSparse) {
 /* ************************************************************************* */
 TEST(PowerMethod, useFactorGraphDense) {
   // Let's make a scalar synchronization graph with 10 nodes
-  GaussianFactorGraph fg;
+  GaussianFactorGraph fg = gtsam::linear::test::example::createDenseGraph();
-  auto model = noiseModel::Unit::Create(1);
-  // Each node has an edge with all the others
-  for (size_t j = 0; j < 10; j++) {
-    fg.add(X(j), -I_1x1, X((j + 1)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 2)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 3)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 4)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 5)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 6)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 7)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 8)%10 ), I_1x1, Vector1::Zero(), model);
-    fg.add(X(j), -I_1x1, X((j + 9)%10 ), I_1x1, Vector1::Zero(), model);
-  }
 
   // Get eigenvalues and eigenvectors with Eigen
   auto L = fg.hessian();