Improvements to Schur template specializations
1. Refactor the python code that generates the template specializations
to remove code duplication.
2. Improved the logic for template specialization selection where
Eigen::Dynamic now serves as a wildcard.
3. Added schur_templates.h/cc which allows querying the set of available
template specializations without instantiating a linear solver.
4. Added Solver::Summary::schur_structre_given and
Solver::Summary::schur_structure_used and expose them in
Solver::Summary::FullReport for better performance debugging.
5. Updated the templates with newer dates and some minor comments cleanup
which lead to the the template specializations to be re-generated.
Change-Id: Iaf3c6f714353597899916c300465da01f151c3de
diff --git a/include/ceres/solver.h b/include/ceres/solver.h
index 0d77d24..45a7c69 100644
--- a/include/ceres/solver.h
+++ b/include/ceres/solver.h
@@ -953,6 +953,19 @@
// parameter blocks.
std::vector<int> linear_solver_ordering_used;
+ // For Schur type linear solvers, this string describes, the
+ // template specialization which was detected in the problem and should be used.
+ std::string schur_structure_given;
+
+ // This is the Schur template specialization that was actually
+ // instantiated and used. The reason this will be different from
+ // schur_structure_given is because the corresponding template
+ // specialization does not exist.
+ //
+ // Template specializations can be added to ceres by editing
+ // internal/ceres/generate_template_specializations.py
+ std::string schur_structure_used;
+
// True if the user asked for inner iterations to be used as part
// of the optimization.
bool inner_iterations_given;
