Release/4.2a4

.github/scripts/python.sh

@@ -83,6 +83,6 @@ cmake $GITHUB_WORKSPACE -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE} \
 make -j2 install
 
 cd $GITHUB_WORKSPACE/build/python
-$PYTHON setup.py install --user --prefix=
+$PYTHON -m pip install --user .
 cd $GITHUB_WORKSPACE/python/gtsam/tests
 $PYTHON -m unittest discover -v

CMakeLists.txt

@@ -11,7 +11,7 @@ endif()
 set (GTSAM_VERSION_MAJOR 4)
 set (GTSAM_VERSION_MINOR 2)
 set (GTSAM_VERSION_PATCH 0)
-set (GTSAM_PRERELEASE_VERSION "a3")
+set (GTSAM_PRERELEASE_VERSION "a4")
 math (EXPR GTSAM_VERSION_NUMERIC "10000 * ${GTSAM_VERSION_MAJOR} + 100 * ${GTSAM_VERSION_MINOR} + ${GTSAM_VERSION_PATCH}")
 
 if (${GTSAM_VERSION_PATCH} EQUAL 0)

doc/Doxyfile.in

@@ -1188,7 +1188,7 @@ USE_MATHJAX            = YES
 # MathJax, but it is strongly recommended to install a local copy of MathJax 
 # before deployment.
 
-MATHJAX_RELPATH        = https://cdn.mathjax.org/mathjax/latest
+# MATHJAX_RELPATH        = https://cdn.mathjax.org/mathjax/latest
 
 # The MATHJAX_EXTENSIONS tag can be used to specify one or MathJax extension 
 # names that should be enabled during MathJax rendering.

doc/gtsam.lyx

@@ -1,5 +1,5 @@
-#LyX 2.2 created this file. For more info see http://www.lyx.org/
-\lyxformat 508
+#LyX 2.3 created this file. For more info see http://www.lyx.org/
+\lyxformat 544
 \begin_document
 \begin_header
 \save_transient_properties true
@@ -62,6 +62,8 @@
 \font_osf false
 \font_sf_scale 100 100
 \font_tt_scale 100 100
+\use_microtype false
+\use_dash_ligatures true
 \graphics default
 \default_output_format default
 \output_sync 0
@@ -91,6 +93,7 @@
 \suppress_date false
 \justification true
 \use_refstyle 0
+\use_minted 0
 \index Index
 \shortcut idx
 \color #008000
@@ -105,7 +108,10 @@
 \tocdepth 3
 \paragraph_separation indent
 \paragraph_indentation default
-\quotes_language english
+\is_math_indent 0
+\math_numbering_side default
+\quotes_style english
+\dynamic_quotes 0
 \papercolumns 1
 \papersides 1
 \paperpagestyle default
@@ -168,6 +174,7 @@ Factor graphs
 \begin_inset CommandInset citation
 LatexCommand citep
 key "Koller09book"
+literal "true"
 
 \end_inset
 
@@ -270,13 +277,15 @@ Let us start with a one-page primer on factor graphs, which in no way replaces
 \begin_inset CommandInset citation
 LatexCommand citet
 key "Kschischang01it"
+literal "true"
 
 \end_inset
 
  and 
 \begin_inset CommandInset citation
 LatexCommand citet
 key "Loeliger04spm"
+literal "true"
 
 \end_inset
 
@@ -1321,6 +1330,7 @@ r in a pre-existing map, or indeed the presence of absence of ceiling lights
 \begin_inset CommandInset citation
 LatexCommand citet
 key "Dellaert99b"
+literal "true"
 
 \end_inset
 
@@ -1542,6 +1552,7 @@ which is done on line 12.
 \begin_inset CommandInset citation
 LatexCommand citealt
 key "Dellaert06ijrr"
+literal "true"
 
 \end_inset
 
@@ -1936,8 +1947,8 @@ reference "fig:CompareMarginals"
 
 \end_inset
 
-, where I show the marginals on position as covariance ellipses that contain
- 68.26% of all probability mass.
+, where I show the marginals on position as 5-sigma covariance ellipses
+ that contain 99.9996% of all probability mass.
  For the odometry marginals, it is immediately apparent from the figure
  that (1) the uncertainty on pose keeps growing, and (2) the uncertainty
  on angular odometry translates into increasing uncertainty on y.
@@ -1992,6 +2003,7 @@ PoseSLAM
 \begin_inset CommandInset citation
 LatexCommand citep
 key "DurrantWhyte06ram"
+literal "true"
 
 \end_inset
 
@@ -2190,9 +2202,9 @@ reference "fig:example"
 \end_inset
 
 , along with covariance ellipses shown in green.
- These covariance ellipses in 2D indicate the marginal over position, over
- all possible orientations, and show the area which contain 68.26% of the
- probability mass (in 1D this would correspond to one standard deviation).
+ These 5-sigma covariance ellipses in 2D indicate the marginal over position,
+ over all possible orientations, and show the area which contain 99.9996%
+ of the probability mass.
  The graph shows in a clear manner that the uncertainty on pose 
 \begin_inset Formula $x_{5}$
 \end_inset
@@ -3076,6 +3088,7 @@ reference "fig:Victoria-1"
 \begin_inset CommandInset citation
 LatexCommand citep
 key "Kaess09ras"
+literal "true"
 
 \end_inset
 
@@ -3088,6 +3101,7 @@ key "Kaess09ras"
 \begin_inset CommandInset citation
 LatexCommand citep
 key "Kaess08tro"
+literal "true"
 
 \end_inset
 
@@ -3355,6 +3369,7 @@ iSAM
 \begin_inset CommandInset citation
 LatexCommand citet
 key "Kaess08tro,Kaess12ijrr"
+literal "true"
 
 \end_inset
 
@@ -3606,6 +3621,7 @@ subgraph preconditioning
 \begin_inset CommandInset citation
 LatexCommand citet
 key "Dellaert10iros,Jian11iccv"
+literal "true"
 
 \end_inset
 
@@ -3638,6 +3654,7 @@ Visual Odometry
 \begin_inset CommandInset citation
 LatexCommand citet
 key "Nister04cvpr2"
+literal "true"
 
 \end_inset
 
@@ -3661,6 +3678,7 @@ Visual SLAM
 \begin_inset CommandInset citation
 LatexCommand citet
 key "Davison03iccv"
+literal "true"
 
 \end_inset
 
@@ -3711,6 +3729,7 @@ Filtering
 \begin_inset CommandInset citation
 LatexCommand citep
 key "Smith87b"
+literal "true"
 
 \end_inset
 

doc/math.lyx

@@ -2668,7 +2668,7 @@ reference "eq:pushforward"
 \begin{eqnarray*}
 \varphi(a)e^{\yhat} & = & \varphi(ae^{\xhat})\\
 a^{-1}e^{\yhat} & = & \left(ae^{\xhat}\right)^{-1}\\
-e^{\yhat} & = & -ae^{\xhat}a^{-1}\\
+e^{\yhat} & = & ae^{-\xhat}a^{-1}\\
 \yhat & = & -\Ad a\xhat
 \end{eqnarray*}
 
@@ -3003,8 +3003,8 @@ between
 \begin_inset Formula 
 \begin{align}
 \varphi(g,h)e^{\yhat} & =\varphi(ge^{\xhat},h)\nonumber \\
-g^{-1}he^{\yhat} & =\left(ge^{\xhat}\right)^{-1}h=-e^{\xhat}g^{-1}h\nonumber \\
-e^{\yhat} & =-\left(h^{-1}g\right)e^{\xhat}\left(h^{-1}g\right)^{-1}=-\exp\Ad{\left(h^{-1}g\right)}\xhat\nonumber \\
+g^{-1}he^{\yhat} & =\left(ge^{\xhat}\right)^{-1}h=e^{-\xhat}g^{-1}h\nonumber \\
+e^{\yhat} & =\left(h^{-1}g\right)e^{-\xhat}\left(h^{-1}g\right)^{-1}=\exp\Ad{\left(h^{-1}g\right)}(-\xhat)\nonumber \\
 \yhat & =-\Ad{\left(h^{-1}g\right)}\xhat=-\Ad{\varphi\left(h,g\right)}\xhat\label{eq:Dbetween1}
 \end{align}
 
@@ -6674,7 +6674,7 @@ One representation of a line is through 2 vectors
 \begin_inset Formula $d$
 \end_inset
 
- points from the orgin to the closest point on the line.
+ points from the origin to the closest point on the line.
 \end_layout
 
 \begin_layout Standard

examples/DiscreteBayesNetExample.cpp

@@ -53,10 +53,9 @@ int main(int argc, char **argv) {
   // Create solver and eliminate
   Ordering ordering;
   ordering += Key(0), Key(1), Key(2), Key(3), Key(4), Key(5), Key(6), Key(7);
-  DiscreteBayesNet::shared_ptr chordal = fg.eliminateSequential(ordering);
 
   // solve
-  auto mpe = chordal->optimize();
+  auto mpe = fg.optimize();
   GTSAM_PRINT(mpe);
 
   // We can also build a Bayes tree (directed junction tree).
@@ -69,14 +68,14 @@ int main(int argc, char **argv) {
   fg.add(Dyspnea, "0 1");
 
   // solve again, now with evidence
-  DiscreteBayesNet::shared_ptr chordal2 = fg.eliminateSequential(ordering);
-  auto mpe2 = chordal2->optimize();
+  auto mpe2 = fg.optimize();
   GTSAM_PRINT(mpe2);
 
   // We can also sample from it
+  DiscreteBayesNet::shared_ptr chordal = fg.eliminateSequential(ordering);
   cout << "\n10 samples:" << endl;
   for (size_t i = 0; i < 10; i++) {
-    auto sample = chordal2->sample();
+    auto sample = chordal->sample();
     GTSAM_PRINT(sample);
   }
   return 0;

examples/DiscreteBayesNet_FG.cpp

@@ -85,7 +85,7 @@ int main(int argc, char **argv) {
         }
 
   // "Most Probable Explanation", i.e., configuration with largest value
-  auto mpe = graph.eliminateSequential()->optimize();
+  auto mpe = graph.optimize();
   cout << "\nMost Probable Explanation (MPE):" << endl;
   print(mpe);
 
@@ -96,8 +96,7 @@ int main(int argc, char **argv) {
   graph.add(Cloudy, "1 0");
 
   // solve again, now with evidence
-  DiscreteBayesNet::shared_ptr chordal = graph.eliminateSequential();
-  auto mpe_with_evidence = chordal->optimize();
+  auto mpe_with_evidence = graph.optimize();
 
   cout << "\nMPE given C=0:" << endl;
   print(mpe_with_evidence);
@@ -110,7 +109,8 @@ int main(int argc, char **argv) {
   cout << "\nP(W=1|C=0):" << marginals.marginalProbabilities(WetGrass)[1]
        << endl;
 
-  // We can also sample from it
+  // We can also sample from the eliminated graph
+  DiscreteBayesNet::shared_ptr chordal = graph.eliminateSequential();
   cout << "\n10 samples:" << endl;
   for (size_t i = 0; i < 10; i++) {
     auto sample = chordal->sample();

examples/HMMExample.cpp

@@ -59,16 +59,16 @@ int main(int argc, char **argv) {
   // Convert to factor graph
   DiscreteFactorGraph factorGraph(hmm);
 
+  // Do max-prodcut
+  auto mpe = factorGraph.optimize();
+  GTSAM_PRINT(mpe);
+
   // Create solver and eliminate
   // This will create a DAG ordered with arrow of time reversed
   DiscreteBayesNet::shared_ptr chordal =
       factorGraph.eliminateSequential(ordering);
   chordal->print("Eliminated");
 
-  // solve
-  auto mpe = chordal->optimize();
-  GTSAM_PRINT(mpe);
-
   // We can also sample from it
   cout << "\n10 samples:" << endl;
   for (size_t k = 0; k < 10; k++) {

examples/UGM_chain.cpp

@@ -68,9 +68,8 @@ int main(int argc, char** argv) {
        << graph.size() << " factors (Unary+Edge).";
 
   // "Decoding", i.e., configuration with largest value
-  // We use sequential variable elimination
-  DiscreteBayesNet::shared_ptr chordal = graph.eliminateSequential();
-  auto optimalDecoding = chordal->optimize();
+  // Uses max-product.
+  auto optimalDecoding = graph.optimize();
   optimalDecoding.print("\nMost Probable Explanation (optimalDecoding)\n");
 
   // "Inference" Computing marginals for each node

examples/UGM_small.cpp

@@ -61,9 +61,8 @@ int main(int argc, char** argv) {
   }
 
   // "Decoding", i.e., configuration with largest value (MPE)
-  // We use sequential variable elimination
-  DiscreteBayesNet::shared_ptr chordal = graph.eliminateSequential();
-  auto optimalDecoding = chordal->optimize();
+  // Uses max-product
+  auto optimalDecoding = graph.optimize();
   GTSAM_PRINT(optimalDecoding);
 
   // "Inference" Computing marginals

gtsam/base/utilities.cpp

@@ -0,0 +1,13 @@
+#include <gtsam/base/utilities.h>
+
+namespace gtsam {
+
+std::string RedirectCout::str() const {
+  return ssBuffer_.str();
+}
+
+RedirectCout::~RedirectCout() {
+  std::cout.rdbuf(coutBuffer_);
+}
+
+}

gtsam/base/utilities.h

@@ -1,5 +1,9 @@
 #pragma once
 
+#include <string>
+#include <iostream>
+#include <sstream>
+
 namespace gtsam {
 /**
  * For Python __str__().
@@ -12,14 +16,10 @@ struct RedirectCout {
   RedirectCout() : ssBuffer_(), coutBuffer_(std::cout.rdbuf(ssBuffer_.rdbuf())) {}
 
   /// return the string
-  std::string str() const {
-    return ssBuffer_.str();
-  }
+  std::string str() const;
 
   /// destructor -- redirect stdout buffer to its original buffer
-  ~RedirectCout() {
-    std::cout.rdbuf(coutBuffer_);
-  }
+  ~RedirectCout();
 
 private:
   std::stringstream ssBuffer_;