commit f7898fba1b92f0e996571b5bfa22a37f5e3644de
author Keir Mierle <mierle@gmail.com> Sat May 05 20:55:08 2012 -0700
committer Keir Mierle <mierle@gmail.com> Sat May 05 20:55:08 2012 -0700
tree 27743595ecfa8990efb19ea5212c214b65d08e85
parent 0a359d6198d257776a8831c3eb98f64ee91cf836

Add a general sparse iterative solver: CGNR

This adds a new LinearOperator which implements symmetric
products of a matrix, and a new CGNR solver to leverage
CG to directly solve the normal equations. This also
includes a block diagonal preconditioner. In experiments
on problem-16, the non-preconditioned version is about
1/5 the speed of SPARSE_SCHUR, and the preconditioned
version using block cholesky is about 20% slower than
SPARSE_SCHUR.

internal/ceres/solver.cc

diff --git a/internal/ceres/solver.cc b/internal/ceres/solver.cc
index c54668e..c45c55f 100644
--- a/internal/ceres/solver.cc
+++ b/internal/ceres/solver.cc
@@ -155,7 +155,8 @@
                           LinearSolverTypeToString(linear_solver_type_given),
                           LinearSolverTypeToString(linear_solver_type_used));
 
-  if (linear_solver_type_given == ITERATIVE_SCHUR) {
+  if (linear_solver_type_given == CONJUGATE_GRADIENTS ||
+      linear_solver_type_given == ITERATIVE_SCHUR) {
     internal::StringAppendF(&report, "Preconditioner      %25s%25s\n",
                             PreconditionerTypeToString(preconditioner_type),
                             PreconditionerTypeToString(preconditioner_type));