Add Covariance documentation to html docs.
Change-Id: I11ddc9f7069964596760c6ea4d85c44312c0a67a
diff --git a/include/ceres/covariance.h b/include/ceres/covariance.h
index 0847357..b70f98b 100644
--- a/include/ceres/covariance.h
+++ b/include/ceres/covariance.h
@@ -129,7 +129,9 @@
//
// The rank deficiency in J can be structural -- columns which are
// always known to be zero or numerical -- depending on the exact
-// values in the Jacobian. This happens when the problem contains
+// values in the Jacobian.
+//
+// Structural rank deficiency occurs when the problem contains
// parameter blocks that are constant. This class correctly handles
// structural rank deficiency like that.
//
@@ -165,21 +167,6 @@
// with indeterminacy. IEEE Transactions on Information Theory 47(5):
// 2017-2028 (2001)
//
-// Speed
-// -----
-//
-// When use_dense_linear_algebra = true, Eigen's JacobiSVD algorithm
-// is used to perform the computations. It is an accurate but slow
-// method and should only be used for small to moderate sized
-// problems.
-//
-// When use_dense_linear_algebra = false, SuiteSparse/CHOLMOD is used
-// to perform the computation. Recent versions of SuiteSparse (>=
-// 4.2.0) provide a much more efficient method for solving for rows of
-// the covariance matrix. Therefore, if you are doing large scale
-// covariance estimation, we strongly recommend using a recent version
-// of SuiteSparse.
-//
// Example Usage
// =============
//
@@ -228,11 +215,18 @@
// estimation of covariance.
int num_threads;
- // Use Eigen's JacobiSVD algorithm to compute the covariance
- // instead of SuiteSparse. This is a very accurate but slow
- // algorithm. The up side is that it can handle numerically rank
- // deficient jacobians. This option only makes sense for small to
+
+ // When use_dense_linear_algebra = true, Eigen's JacobiSVD
+ // algorithm is used to perform the computations. It is an
+ // accurate but slow method and should only be used for small to
// moderate sized problems.
+ //
+ // When use_dense_linear_algebra = false, SuiteSparse/CHOLMOD is
+ // used to perform the computation. Recent versions of SuiteSparse
+ // (>= 4.2.0) provide a much more efficient method for solving for
+ // rows of the covariance matrix. Therefore, if you are doing
+ // large scale covariance estimation, we strongly recommend using
+ // a recent version of SuiteSparse.
bool use_dense_linear_algebra;
// If the Jacobian matrix is near singular, then inverting J'J
