- Boost-users - lists.boost.org

Boost for GPUs
by Blesson Varghese 17 Apr '13

17 Apr '13

Hi All I am new to using Boost libraries. I have written a function that makes use of the boost for computing the inverse of the cummulative distribution function of the beta distribution (quantile) in C++ which works fine. Now I would like to use the same function on the GPU, but I cannot call a host function from device. Are there any alternatives to use boost functions on GPU? Looking forward to your reply. Many thanks and kind regards Blesson

4 5

[timer] Error linking boost::chrono::steady_clock::now
by Michael Powell 16 Apr '13

16 Apr '13

Hello, I get the following linker error. Description Resource Path Location Type undefined reference to `boost::chrono::steady_clock::now()' detector line 0 C/C++ Problem I am using Boost 1.53. A little unclear what the docs are talking about. If I can go header only, and if it is advantageous to do so, I'd like to. http://www.boost.org/doc/libs/1_53_0/doc/html/chrono/users_guide.html#chron… However, not sure what else we're talking about here, when it talks about building boost with buggy header only support (?). Regards, Michael Powell

3 4

Re: [Boost-users] [graph] dijkstra additional constrains that will stop the search (ex. distance from source)
by The Maschine 16 Apr '13

16 Apr '13

:) My description of the problem is hopeless. The boost::dijkstra_shortest_paths(m_ugraph, *vertex_iterator_begin, boost:: predecessor_map(predecessorMap).distance_map(distanceMap)); currently works ok. Finds the shortest path based on the "edge_weight" (this is what I need). The addition I need: I need the search for the shortest path to stop if outside a radius. (my graph is a street network). lets say max "travel distance" = 4 Source -> V1(length=1) -> V2(length=3) -> V3(length=2) -> V4(length=1) -> V5 (length=1) ->VEnd The search should end on V3 and V3 should be "unreachable" from the source (V4,V5 and Vend should be unreachable too) The distance_map should have the distances for V1 and V2 based on the "edge_weight". This explanation seems to be better. Thanks Tasos

2 1

How to print boost::exceptions to ostream
by Michael Powell 16 Apr '13

16 Apr '13

Hello, Basically like the subject says. I need a way to print boost::exceptions through an ostream. Or std::exception if that's a better more portable route to take. Suggestions? Regards, Michael Powell

2 1

Need good semi-high-res timer
by Michael Powell 16 Apr '13

16 Apr '13

Hello, I am looking for a semi-high-res timer solution. Ideally, should be clock-agnostic, shouldn't see non-deterministic behavior in the face of NTP or GPS system or local time updates. Thus far attempts to use boost/timer/timer.hpp have been frustrated by what seems like the common occurrence linker issues. I've defined to use header only, but that doesn't seem to have fixed it. Or possibly I need to rebuild boost with that option? Or something along these lines. I'd like to use boost if that feature is available, and avoid clock-bound timers if at all possible. Recommendations? Thanks much. Regards, Michael Powell

1 0

Re: [Boost-users] boost asio daemon
by Marat Abrarov 16 Apr '13

16 Apr '13

Hi, Philipp. > io::io_service l_ioservice; > io::signal_set l_signals(l_ioservice, SIGINT, SIGTERM); > l_signals.async_wait(daemonhandler); // * > l_ioservice.run(); // ** > > void daemonhandler( const boost::system::Error_code& err ) { > if (!err) > do some log > else > do daemon work > } > > My question is, is the daemonhandler function a function that is called > only if the daemon is stopped or that is run during the daemon is active / > working? According to Boost.Asio docs (http://www.boost.org/doc/libs/1_53_0/doc/html/boost_asio/overview/core/basi cs.html) all callbacks (handlers) are executed (handler::operator() is called) within the threads running asio::io_service::run() (run_one/poll/poll_one). So at the execution point of the daemonhandler function at least one thread of daemon (thread of the daemon's process) runs. If the above code is the only place that executes asio::io_service::run() (see **) then daemonhandler function will be called within the same thread. And if (*) is the only pending async operation then continuation of execution of asio::io_service::run() depends on daemonhandler starting additional async operations on the same instance of asio::io_service (on l_ioservice). Regards, Marat Abrarov.

2 1

[timer] Examples do not compile
by Michael Powell 16 Apr '13

16 Apr '13

Hello, I am trying to setup a "simple" elapsed timer abstraction to hide the boost::timer::cpu_timer concern. However, I am finding that the usage example illustrated here does not compile, or I am not including something I need to include (more likely). http://www.boost.org/doc/libs/1_53_0/libs/timer/doc/cpu_timers.html#using-c… Might there be more complete working examples I can draw from? Thank you... Regards, Michael Powell

1 1

Re: [Boost-users] [graph] dijkstra additional constrains that will stop the search (ex. distance from source)
by The Maschine 16 Apr '13

16 Apr '13

Hi, Not really! (Im not that advance yet) Let me explain. 1) Every vertex in the graph has each own m_length value which is the 'length' of that node only. 2) from the "source" of the search, I want the search to extent to only the node in the radius from the source. So that "radius" cutoff should by something like adding all "m_length" of the route to the current node. At the moment my search only has a weight_map of the "Edge Weights" not some sort of "Vertex Weight". How should I proceed? Is this clear? Sorry for the confusion. Thanks Tasos > I had assumed you were using m_length as the distance map in Dijkstra's > algorithm, and so that would be the total distance you want to compare to. > Is that what you intended? If not, what are you using as your distance > map? > > -- Jeremiah Willcock > >

2 1

program_options allow_unregistered and *nix style
by Michael Powell 16 Apr '13

16 Apr '13

Hello, I would like to find a way to allow_unregistered program_options, as well as specify cls::unix_style ^ cls::allow_short, where cls is po::command_line_style. The code I've got here doesn't seem to like unrecognized ones: //vis_ are my visible options, vars_ is my variable map. po::store(parse_command_line(argc, argv, vis_, cls::unix_style ^ cls::allow_short), vars_); po::notify(vars_); And then a slight variation with apparently a way to do this (untested), how to specify that cls? // http://stackoverflow.com/questions/10178525/how-to-handle-unsolicited-param… po::parsed_options parsed_ = po::command_line_parser(argc, argv) .options(vis_).allow_unregistered().run(); Possible? Thank ye... Regards, Michael Powell

2 4

Re: [Boost-users] [graph] dijkstra additional constrains that will stop the search (ex. distance from source)
by The Maschine 16 Apr '13

16 Apr '13

Hi Jeremiah, As far as I understand this will examine only the "local" value (node length for example) against the threshold. In my path search I need to be able to find the "total length/distance" from the source to the vertex in question. Adding up the lengths (of the nodes that the path is passing) from source to current vertex and then tested against the threshold. Am I clear or my description is wrong? Thanks a lot, Tasos > Here's the basic idea (not tested); use this as your Dijkstra visitor, and > change the part labeled "threshold" to your threshold computation: > > struct threshold_visitor: boost::default_dijkstra_visitor { > double threshold; > > threshold_visitor(double threshold): threshold(threshold) {} > > struct hit_threshold {}; > > void examine_vertex(vertex_descriptor v, const unGraph& g) { > if (get(&VertexProperties::m_ref, g, v) > threshold) > throw hit_threshold(); > } > }; > > Remember to wrap your call to Dijkstra's algorithm in a try-catch block > for threshold_visitor::hit_threshold. > > -- Jeremiah Willcock > >

2 1