- Boost-users - lists.preview.boost.org

[Graph] [BGL] trying to modify brandes betweenness impl
by The Maschine 06 Jan '15

06 Jan '15

Hello all and happy new year! Im trying to modify the brandes centrality in order to analyse only a subset of the graph based on a cutoff distance calculated by some other edge weight than the one used for the brandes call. What I have done so far is to add a shortest path search call before the main path search inside brandes_betweenness_centrality_impl. The first search uses the edge::m_metric_weight and constructs the distance map from 's' based on that weight. Then I feed the metric distance map in the 'brandes special visitor' and I check inside visitor's examine_vertex. Reminder that the brandes call uses the m_special_weight. While this works as a principle I have no idea how people with experience would do it! A problem with this is that while the metric cutoff can be small you still run both searches to the max (as if there is no cutoff). Any help on how to change the first 'dijkstra_shortest_paths' so that after reaching the cutoff the search stops gracefully and leaves the rest of the 'distance_map' to infinity? Thanks a lot and please let me know if something need more clarification. ############### struct EdgeProperties { double m_special_weight; double m_metric_distance_weight; }; typedef boost::adjacency_list<boost::vecS, boost::vecS, boost::undirectedS, VertexProperties, EdgeProperties> unGraph; // test my_brandes_dijkstra_visitor // template<typename Graph, typename WeightMap, typename IncomingMap, typename DistanceMap, typename PathCountMap> struct my_brandes_dijkstra_visitor : public bfs_visitor<> { typedef typename graph_traits<Graph>::vertex_descriptor vertex_descriptor; typedef typename graph_traits<Graph>::edge_descriptor edge_descriptor; my_brandes_dijkstra_visitor(std::stack<vertex_descriptor>& iordered_vertices, WeightMap iweight, IncomingMap iincoming, DistanceMap idistance, PathCountMap ipath_count, std::vector<float> c_dists, float coff) : ordered_vertices(iordered_vertices), weight(iweight), incoming(iincoming), distance(idistance), path_count(ipath_count), cutoff_dists(c_dists), cutoff(coff) { } /** * Whenever an edge e = (v, w) is relaxed, the incoming edge list * for w is set to {(v, w)} and the shortest path count of w is set to * the number of paths that reach {v}. */ void edge_relaxed(edge_descriptor e, const Graph& g) { vertex_descriptor v = source(e, g), w = target(e, g); incoming[w].clear(); incoming[w].push_back(e); put(path_count, w, get(path_count, v)); } /** * If an edge e = (v, w) was not relaxed, it may still be the case * that we've found more equally-short paths, so include {(v, w)} in the * incoming edges of w and add all of the shortest paths to v to the * shortest path count of w. */ void edge_not_relaxed(edge_descriptor e, const Graph& g) { typedef typename property_traits<WeightMap>::value_type weight_type; typedef typename property_traits<DistanceMap>::value_type distance_type; vertex_descriptor v = source(e, g), w = target(e, g); distance_type d_v = get(distance, v), d_w = get(distance, w); weight_type w_e = get(weight, e); closed_plus<distance_type> combine; if (d_w == combine(d_v, w_e)) { put(path_count, w, get(path_count, w) + get(path_count, v)); incoming[w].push_back(e); } } /// Keep track of vertices as they are reached void examine_vertex(vertex_descriptor w, const Graph&) { // add only if this is within the 'cutoff'. cutoff_dists holds // the metric distances from s. if( cutoff_dists[w] < cutoff ) ordered_vertices.push(w); } private: std::stack<vertex_descriptor>& ordered_vertices; WeightMap weight; IncomingMap incoming; DistanceMap distance; PathCountMap path_count; std::vector<float> cutoff_dists; float cutoff; }; //////////////////////////////////////////////////////////////////// template<typename WeightMap> struct my_brandes_dijkstra_shortest_paths { my_brandes_dijkstra_shortest_paths(WeightMap iweight_map) : weight_map(iweight_map) { } template<typename Graph, typename IncomingMap, typename DistanceMap, typename PathCountMap, typename VertexIndexMap> void operator()(Graph& g, typename graph_traits<Graph>::vertex_descriptor s, std::stack<typename graph_traits<Graph>::vertex_descriptor>& ov, IncomingMap incoming, DistanceMap distance, PathCountMap path_count, VertexIndexMap vertex_index , std::vector<float> c_dists, float coff) // modified cutoff vector and value { typedef my_brandes_dijkstra_visitor<Graph, WeightMap, IncomingMap, DistanceMap, PathCountMap> visitor_type; visitor_type visitor(ov, weight_map, incoming, distance, path_count,c_dists,coff); // modified cutoff vector and value dijkstra_shortest_paths(g, s, boost::weight_map(weight_map) .vertex_index_map(vertex_index) .distance_map(distance) .visitor(visitor)); } private: WeightMap weight_map; }; ////////////////////////////// template<typename Graph, typename CentralityMap, typename EdgeCentralityMap, typename IncomingMap, typename DistanceMap, typename DependencyMap, typename PathCountMap, typename VertexIndexMap, typename ShortestPaths> void my_brandes_betweenness_centrality_impl(const Graph& g, CentralityMap centrality, // C_B EdgeCentralityMap edge_centrality_map, IncomingMap, //incoming, // P // OpenMP mod DistanceMap, //distance,// d // OpenMP mod DependencyMap, //dependency,// delta // OpenMP mod PathCountMap, //path_count,// sigma // OpenMP mod VertexIndexMap vertex_index, ShortestPaths shortest_paths) { typedef typename graph_traits<Graph>::vertex_iterator vertex_iterator; typedef typename graph_traits<Graph>::edge_iterator edge_iterator; typedef typename graph_traits<Graph>::vertex_descriptor vertex_descriptor; // Initialize centrality init_centrality_map(vertices(g), centrality); init_centrality_map(edges(g), edge_centrality_map); int i, N = num_vertices(g); #pragma omp parallel for default(shared) private(i) schedule(dynamic) for (i = 0; i < N; ++i) { typename graph_traits<Graph>::vertex_descriptor s = vertex(i, g); if (s == graph_traits<Graph>::null_vertex()) continue; // First we build a map of all distances from 's' based // on m_metric_distance_weight. typedef typename boost::property_map < Graph, boost::vertex_index_t >::type CIndexMap; typedef typename boost::iterator_property_map < vertex_descriptor*, CIndexMap, vertex_descriptor, vertex_descriptor& > CPredecessorMap; typedef typename boost::iterator_property_map < float*, CIndexMap, float, float& > CDistanceMap; float co = 2200.0f; std::vector<vertex_descriptor> cut_predecessors(boost::num_vertices(g)); // To store parents std::vector<float> cut_distances(boost::num_vertices(g), std::numeric_limits<double>::max() ); // To store distances CIndexMap cut_indexMap = boost::get(boost::vertex_index, g); CPredecessorMap cut_predecessorMap(&cut_predecessors[0], cut_indexMap); CDistanceMap cut_distanceMap(&cut_distances[0], cut_indexMap); // boost::dijkstra_shortest_paths(g, s, boost::weight_map(get(&EdgeProperties::m_metric_distance_weight, g)).predecessor_map(cut_predecessorMap).distance_map(cut_distanceMap)); /////////// // Local storage for OpenMP std::stack<vertex_descriptor> ordered_vertices; vector_property_map<typename property_traits<IncomingMap>:: value_type,VertexIndexMap> incoming(vertex_index); vector_property_map<typename property_traits<DistanceMap>:: value_type,VertexIndexMap> distance(vertex_index); vector_property_map<typename property_traits<DependencyMap>:: value_type,VertexIndexMap> dependency(vertex_index); vector_property_map<typename property_traits<PathCountMap>:: value_type,VertexIndexMap> path_count(vertex_index); // Initialize for this iteration vertex_iterator w, w_end; for (tie(w, w_end) = vertices(g); w != w_end; ++w) { incoming[*w].clear(); put(path_count, *w, 0); put(dependency, *w, 0); } put(path_count, s, 1); // co - cutoff added. b_centrality is calculated based on // Edge::m_special_weight while we are testing // to be inside the 'cutoff' that is based on the // Edge::m_metric_weight. shortest_paths(g, s, ordered_vertices, incoming, distance, path_count, vertex_index, cut_distances, co); while (!ordered_vertices.empty()) { vertex_descriptor u = ordered_vertices.top(); ordered_vertices.pop(); typedef typename property_traits<IncomingMap>::value_type incoming_type; typedef typename incoming_type::iterator incoming_iterator; typedef typename property_traits<DependencyMap>::value_type dependency_type; for (incoming_iterator vw = incoming[u].begin(); vw != incoming[u].end(); ++vw) { vertex_descriptor v = source(*vw, g); dependency_type factor = dependency_type(get(path_count, v)) / dependency_type(get(path_count, u)); factor *= (dependency_type(1) + get(dependency, u)); put(dependency, v, get(dependency, v) + factor); update_centrality(edge_centrality_map, *vw, factor); } if (u != s) { #pragma omp critical(globalupdate) { update_centrality(centrality, u, get(dependency, u)); } } } } typedef typename graph_traits<Graph>::directed_category directed_category; const bool is_undirected = is_convertible<directed_category*, undirected_tag*>::value; if (is_undirected) { divide_centrality_by_two(vertices(g), centrality); divide_centrality_by_two(edges(g), edge_centrality_map); } } //

1 0

pass asio socket to another class
by cap 05 Jan '15

05 Jan '15

I have two classes SslHandshake and SslRequest with according to names functionality. After client and server handshakes are done in the SslHandshake, I pass both sockets to SslRequest and start transferring data from server to client. 1. If I is not waiting for finishing of transferring, instance of SslHandshake will be destroyed automatically (in the ~SslHandshake I will destroy both socket objects and transferring in the SslRequest will be aborted). 2. If I start transferring with waiting, after a few iterations SslRequest::ReadSocketOut will be called successfully, but the readSocketHandle will not be called (so, will be called but with timeout error). So, I need to handshake in one class and transfer data into another class. How to prevent destroying instance of SslHandshake until data transferring will be done into SslRequest or how to fix socket reading without callback calling after a few iterations? void SslHandshake::HandshakeInHandle(const bSystem::error_code &error) { TRY { CHECK_BOOST_ERROR { LOG_MSG_DEBUG_KEYWORD(KEYWORD_INCOMING, << "Handshake done"); // --- Transfer with waiting of mutex //IoServicePtr servicePointer((IoService *)&m_SocketIn->get_io_service()); //boost::thread serviceThread(boost::bind(&IoService::IO_SERVICE_THREAD_VOID, servicePointer)); //SslRequest::SharedPtr sslRequest(new SslRequest(m_SocketIn, m_SocketOut)); //sslRequest->Transfer(); SslRequest::SocketPtr socketIn(m_SocketIn); SslRequest::SocketPtr socketOut(m_SocketOut); // --- Transfer with waiting of thread //IoServicePtr servicePointer((IoService *)&m_SocketIn->get_io_service()); //servicePointer->SessionIndex = IndexThis; //boost::thread serviceThread(boost::bind(&IoService::IO_SERVICE_THREAD_VOID, servicePointer)); //SslRequest::SharedPtr request = SslRequest::CreateInstance(m_SocketIn, m_SocketOut); //request->Transfer(); //serviceThread.join(); // --- Transfer with waiting of IO_SERVICE_THREAD_VOID IoServicePtr servicePointer((IoService *)&m_SocketIn->get_io_service()); servicePointer->SessionIndex = IndexThis; SslRequest::SharedPtr request = SslRequest::CreateInstance(/*m_SocketIn, m_SocketOut*/socketIn, socketOut INDEX_THIS_PASS); request->Transfer(); servicePointer->IO_SERVICE_THREAD_VOID(); } } CATCH_LOG_STD CATCH_LOG_ANY } void SslRequest::ReadSocketOut(TransferringBuffer &buffer, SimpleSocketCallback readSocketHandle) { bAsIO::async_read( *m_SocketOut, bAsIO::buffer(buffer), boost::bind( readSocketHandle, SHARED_FROM_THIS, baPlaceholders::error, baPlaceholders::bytes_transferred)); } -- View this message in context: http://boost.2283326.n4.nabble.com/pass-asio-socket-to-another-class-tp4670… Sent from the Boost - Users mailing list archive at Nabble.com.

2 3

Boost.DateTime library to synthesise time-series data?
by Riskybiz 05 Jan '15

05 Jan '15

Dear boost-users, It's my first time trying this and hope that someone may be able to enlighten me about which aspects of the Boost.DateTime library will help solve this problem... I have some time-series data which is time stamped with boost::posix_time::ptime values. From this base-series I would like to synthesise higher interval time-series. For example, a base-series of 1 second time-series data could be used to synthesise a time-series of any greater interval; 2 seconds, 5 seconds, 86400 seconds (1 day) etc. Or 1 minute base-series data could be used to synthesise, 2 min, 15min, 60 min time-series etc. It is entirely possible that any given base-series time point could be absent i.e. there could be time gaps in the base-series. I have an idea to use boost::posix_time::time_period to create regularly temporally spaced 'buckets' of the required time interval; into which the base-series data elements would accumulate and thus allow the synthesising of the higher interval time-series. Each 'bucket' represents one higher interval time-series data element. It is important that the 'buckets' are aligned with natural multiples of the time interval for which they are accumulating. So for example a 10 second synthesised time-series would have boost::posix_time::time_period 'buckets' like so at: 1-10, 11-20, 21-30, 31-40, 41-50 & 51-60 seconds time_periods. The first data element of the base-series could fall at any time, it could be on the 7th second of a minute, equally it could be on the 59th minute of an hour. Somehow the first boost::posix_time::time_period 'bucket' for a particular synthesised interval must be set such that it spans the correct period of time and ends on a multiple of the time-series interval which is being synthesised. Put simply, that first base-series data element needs to fall into a correctly aligned time_period 'bucket'. Using Boost.Date_Time how could the boost::posix_time::ptime of the first data element of the base-series be examined and the first time_period 'bucket' be correctly aligned to the relevant begin and end times for the synthesised time-series? I envisage subsequent 'buckets' will be created by incrementing forward in time from the datum of the first. With grateful thanks for any assistance, Riskybiz.

2 1

How to build Boost with VC++ 2013 to target XP/2003 platform?
by Caroline Beltran 01 Jan '15

01 Jan '15

Programs written in VC++ 2012 or 2013 will only run on Windows XP or Windows 2003 when you choose the 'Windows XP (v120_xp)' toolset option in the IDE. Is there some way to set this option from Bjam? P.S. A Stack Overflow user indicated that he got Boost to build under VC++ 2012. I tried his solution under VC++ 2013 but it did not work for some reason completely unknown to me. http://stackoverflow.com/questions/27695187/modify-visual-c-vcvars32-bat-to… Apparently, VC++ 2012/2013 use the Windows 8 SDK and the solution above causes you to use the Windows 7 SDK which is compatible with XP and Windows 2003. Please help me

2 2

[boost] [compute] Review results
by Antony Polukhin 01 Jan '15

01 Jan '15

Hi all, Extended review period for Boost.Compute library has ended. Here are the review votes for inclusion (+ means include, - means do not include, +/- means conditional include or "probably include, not quite sure"): + Denis Demidov + Sebastian Schaetz + Pavan Yalamanchili - Gruenke,Matt (+ if more backends will be supported) + Asbjørn (mini-review/comments from a library user) + Thomas M (a lot of suggestions) + Paul A. Bristow + Belcourt, Kenneth + John Bytheway +/- Yiannis Papadopoulos (there's already Bolt/Thrust for beginner and pure OpenCL for pros) +/- Vicente J. Botet Escriba (better docs and more performance tests) +/- Jason Newton (there was no explicit answer to the question "shuld be this library accepted into Boost?" but the review was positive in general) To sum up: Compute library becomes part of the Boost. Great thanks to all for the reviews and comments! Huge work was done and the issues raised are essential for library success. P.S.: Happy new year everyone! P.P.S.: Hope I did not miss some review! In case of any error, please let me know! -- Best regards, Antony Polukhin

1 0

Writing custom allocator for use with extended boost vector
by Olumide 31 Dec '14

31 Dec '14

I am considering using boost::container::vector as an alternative to std::vector because std::vector moves when its capacity is exceeded. This is a problem because I am using a lean stack-like allocator that acquires a pool of memory from the heap and manages it my incrementing or decrementing a pointer to the available memory; and automatically moving the vector in the pool is a big no-no for several reasons chief of which is that such moves invalidate the above pointer. My allocator does not use a header block in order to delineate subsequent allocations. This approach works because my application is recursive i.e. objects are allocated and deallocated in a strict LIFO order. Furthermore allocations are typically small (but not constant size) and and are routinely smaller than the size of the canonical header block consisting of a forward and backward pointer. To use such a header block would IMHO be a waste. I'd like to port this allocator to for use with boost::container::vector but I'm not sure entirely sure how much of the boost::container::allocator interface that I need to implement, or if perchance there is already an allocator that does what I am trying to do. Please advise. Regards, - Olumide

1 0

nowide
by Colin Humphries 30 Dec '14

30 Dec '14

Artyom, Please ignore my request for wifstream and wofstream constructors. I see now that these are not needed: windows accepts UTF-16 filenames so of course I just needed to use widen. I am however surprised that your fstream constructors do not directly accept std::string? I feel that the required .c_str() call should be incorporated. Regards, Colin Humphries

1 0

nowide
by Colin Humphries 29 Dec '14

29 Dec '14

Hello Artyom, Please provide wifstream and wofstream constructors. I realise this request may seem contrary to the spirit of your very useful library but wide streams are required by boost::serialize to output to XML (i.e. UTF-8) (Yes, I agree it might be more sensible to update Boost::Serialize). Regards, Colin Humphries

1 0

[filesystem] ISO TS 18822, Technical Specification - File System
by Beman Dawes 29 Dec '14

29 Dec '14

The filesystem TS has been unanimously approved by ISO. 32 countries voted. The document has been forwarded to the ISO Central Secretariat for publication. The current working draft, which is technically the same as the TS but may have editorial corrections, is available at http://cplusplus.github.io/filesystem-ts/working-draft.html Many thanks to all the Boost folks who contributed so much to the Boost library that became the ISO TS! The C++ standards committee Filesystem Study Group will continue to operate, and will issue further TSes as further proposals dictate, and will continue to publish a working paper as issues for the TS are corrected. Happy Boost New Year, --Beman

1 0

[Serialize] Conflict between Serialize and PIMPL?
by Merrill Cornish 28 Dec '14

28 Dec '14

In a normal PIMPL structure, the visible class is responsible for instantiating the private implementation class when the visible class is instantiated. However, when serialization is restoring objects, /it/ thinks it is responsible for instantiating all classes, including the private implementation classes of a PIMPL structure. Who wins? Do I end up with /two/ instances of the private implementation class, one of which is just hanging in mid-air? If you need a more solid background, here is what I'm doing: _Visible class header_: struct Foo_Private; class Foo { public: Foo(); . . ./other//stuff/ . . . . . . /data member-free serialization//code/ . . . private: *const std::unique_ptr< Foo_Private() > d_ptr**;* }; _Visible class code_: Foo::Foo() : *d_ptr(std::make_unique< Foo_Private >())* {...code to initialize Foo_Private data members...} _Private Implementation class header_: struct Foo_Private{ Foo_Private(); . . . /data member declarations/ . . . . . . /data member serialization code/ . . . }; Private implementation class has no code file. It is a header-only class. One alternative I thought of was to /not/ serialize the private implementation class, even though it contains all of the data member that need serializing. Instead, if the private class has data members, say, bar and baz, then the /visible/ class would have a serialize() function that "reached into" the private object using its d_ptr: template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & d_ptr->bar; ar & d_ptr->baz; } So, the visible class (which has no data members of its own) is serialized using data extracted from/inserted into the private class object. Would serialization-by-proxy work?

2 1