Class boost::adjacency_list< OutEdgeListS, distributedS< ProcessGroup, InVertexListS, InDistribution >, DirectedS, VertexProperty, EdgeProperty, GraphProperty, EdgeListS >

Determine precisely what we should do with nonlocal edges in undirected graphs. Our parallelization of certain algorithms relies on the ability to access edge property maps immediately (e.g., edge_weight_t), so it may be necessary to duplicate the edge properties in both processes (but then we need some form of coherence protocol).

What does the user do if property_reduce doesn't do the right thing?

Member boost::adjacency_list< OutEdgeListS, distributedS< ProcessGroup, InVertexListS, InDistribution >, DirectedS, VertexProperty, EdgeProperty, GraphProperty, EdgeListS >::add_remote_edge (const msg_nonlocal_edge_data &, processor_id_type, directedS)

There is a potential problem here, where we could unintentionally allow duplicate edges in undirected graphs because the same edge is added on two different processors simultaneously. It's not an issue now, because we require that the graph allow parallel edges. Once we do support containers such as setS or hash_setS that disallow parallel edges we will need to deal with this.

Member boost::detail::parallel::operator== (const edge_descriptor< Edge > &e1, const edge_descriptor< Edge > &e2)

need edge_descriptor to know if it is undirected so we can compare both ways.

Class boost::exterior_property< Graph, Key, Value >

This does not currently support the ability to define exterior properties for graph types that do not model the IndexGraph concepts. A solution should not be especially difficult, but will require an extension of type traits to affect the type selection.

Class boost::geometry::strategy::transform::rotate_transformer< DegreeOrRadian, CalculationType, Dimension1, Dimension2 >

The 3D version will not compile.

Class boost::graph::distributed::mpi_process_group::outgoing_messages

Evaluate whether we should use a deque instance, which would reduce could reduce the cost of "sending" messages but increases the time spent in the synchronization step.

Class boost::labeled_graph< Graph, Label, Selector >

This class is somewhat redundant for adjacency_list<*, vecS> if the intended label is an unsigned int (and perhaps some other cases), but it does avoid some weird ambiguities (i.e. adding a vertex with a label that does not match its target index).

This needs to be reconciled with the named_graph, but since there is no documentation or examples, its not going to happen.

Member boost::lengauer_tarjan_dominator_tree_without_dfs (const Graph &g, const typename graph_traits< Graph >::vertex_descriptor &entry, const IndexMap &, TimeMap dfnumMap, PredMap parentMap, VertexVector &verticesByDFNum, DomTreePredMap domTreePredMap)

: Optimization in Finding Dominators in Practice, Loukas Georgiadis

Member boost::numeric::ublas::blas_3::hr2k (M1 &m1, const T1 &t1, const T2 &t2, const M2 &m2, const M3 &m3)

use opb_prod()

Member boost::numeric::ublas::blas_3::hrk (M1 &m1, const T1 &t1, const T2 &t2, const M2 &m2)

use opb_prod()

Member boost::numeric::ublas::blas_3::sr2k (M1 &m1, const T1 &t1, const T2 &t2, const M2 &m2, const M3 &m3)

use opb_prod()

Member boost::numeric::ublas::blas_3::srk (M1 &m1, const T1 &t1, const T2 &t2, const M2 &m2)

use opb_prod()

Class boost::numeric::ublas::matrix_expression_assigner< E, Fill_Policy, Traverse_Policy >

Add examples link

Class boost::numeric::ublas::matrix_move_to_manip< T >

Doxygen has some problems with similar template functions. Correct that.

Member boost::numeric::ublas::move ()

Doxygen has some problems with similar template functions. Correct that.

Member boost::numeric::ublas::move_to (T i, T j)

Doxygen has some problems with similar template functions. Correct that.

Member boost::numeric::ublas::operator<<= (matrix_expression< E > &me, const index_manipulator< T > &ta)

Add examples link

Member boost::numeric::ublas::operator<<= (vector_expression< E > &v, const index_manipulator< T > &nv)

Add examples link

Member boost::numeric::ublas::operator<<= (vector_expression< E > &v, const typename E::value_type &val)

Add examples link

Member boost::numeric::ublas::operator<<= (vector_expression< E1 > &v, const vector_expression< E2 > &ve)

Add examples link

Member boost::numeric::ublas::operator<<= (matrix_expression< E1 > &me1, const matrix_expression< E2 > &me2)

Add examples link

Member boost::numeric::ublas::operator<<= (matrix_expression< E > &me, traverse_policy_wrapper< T >)

Add examples link

Member boost::numeric::ublas::operator<<= (matrix_expression< E > &me, fill_policy_wrapper< T >)

Add examples link

Member boost::numeric::ublas::operator<<= (vector_expression< E > &v, fill_policy_wrapper< T >)

Add examples link

Member boost::numeric::ublas::operator<<= (matrix_expression< E > &me, const typename E::value_type &val)

Add examples link

Member boost::numeric::ublas::operator<<= (matrix_expression< E1 > &me, const vector_expression< E2 > &ve)

Add examples link

Class boost::numeric::ublas::static_matrix_move_manip< I, J >

Doxygen has some problems with similar template functions. Correct that.

Class boost::numeric::ublas::static_matrix_move_to_manip< I, J >

Doxygen has some problems with similar template functions. Correct that.

Class boost::numeric::ublas::static_vector_move_manip< I >

Doxygen has some problems with similar template functions. Correct that.

Class boost::numeric::ublas::traverse_policy::by_column_policy< Wrap >

Add examples link

Class boost::numeric::ublas::traverse_policy::by_row_policy< Wrap >

Add examples link

Member boost::numeric::ublas::vector< T, A >::vector (size_type, const array_type &data)

remove this definition because size is not used

Class boost::numeric::ublas::vector_expression_assigner< E, Fill_Policy >

Add examples link

Class boost::parallel::caching_property_map< PropertyMap >

Find a better name for caching_property_map