Add dynamic_sparsity option.
The standard sparse normal Cholesky solver assumes a fixed
sparsity pattern which is useful for a large number of problems
presented to Ceres. However, some problems are symbolically dense
but numerically sparse i.e. each residual is a function of a
large number of parameters but at any given state the residual
only depends on a sparse subset of them. For these class of
problems it is faster to re-analyse the sparsity pattern of the
jacobian at each iteration of the non-linear optimisation instead
of including all of the zero entries in the step computation.
The proposed solution adds the dynamic_sparsity option which can
be used with SPARSE_NORMAL_CHOLESKY. A
DynamicCompressedRowSparseMatrix type (which extends
CompressedRowSparseMatrix) has been introduced which allows
dynamic addition and removal of elements. A Finalize method is
provided which then consolidates the matrix so that it can be
used in place of a regular CompressedRowSparseMatrix. An
associated jacobian writer has also been provided.
Changes that were required to make this extension were adding the
SetMaxNumNonZeros method to CompressedRowSparseMatrix and adding
a JacobianFinalizer template parameter to the ProgramEvaluator.
Change-Id: Ia5a8a9523fdae8d5b027bc35e70b4611ec2a8d01
diff --git a/internal/ceres/linear_solver.h b/internal/ceres/linear_solver.h
index 8737c7c..f091bc5 100644
--- a/internal/ceres/linear_solver.h
+++ b/internal/ceres/linear_solver.h
@@ -98,6 +98,7 @@
dense_linear_algebra_library_type(EIGEN),
sparse_linear_algebra_library_type(SUITE_SPARSE),
use_postordering(false),
+ dynamic_sparsity(false),
min_num_iterations(1),
max_num_iterations(1),
num_threads(1),
@@ -115,6 +116,7 @@
// See solver.h for information about this flag.
bool use_postordering;
+ bool dynamic_sparsity;
// Number of internal iterations that the solver uses. This
// parameter only makes sense for iterative solvers like CG.
