June 2009 - Boost-users - lists.boost.org

Random library, discrete distribution
by Kevin Martin 18 Aug '09

18 Aug '09

I made a post sometime last year about requiring a distribution that could generate enums with specified probabilities. I find I am regularly requiring a discrete distribution like this - where I specify the probabilities. Am I the only one who requires it, or would it be useful to a wider audience? I'm just wondering whether to submit a feature request for it or not? Thanks, Kevin Martin This is what I'm using at the moment: template<class IntType = int> class discrete_distribution { public: typedef boost::uniform_int<>::input_type input_type; typedef IntType result_type; template<size_t N> discrete_distribution(const boost::array<int, N> &outputs, const boost::array<int, N> &weights) : outputs_(outputs.begin(), outputs.end()), weights_(weights.begin(), weights.end()) { for(int i=1;i!=weights_.size();++i) weights_[i] += weights_[i-1]; } inline void reset() {} template<class Engine> result_type operator()(Engine& eng) { boost::uniform_int<> ui(0, weights_.back()-1); boost::variate_generator<Engine&, boost::uniform_int<> > vg(eng, ui); int idx = std::lower_bound(weights_.begin(), weights_.end(), vg()) - weights_.begin(); return outputs_[idx]; } private: std::vector<int> outputs_; std::vector<int> weights_; };

3 11

Re: [Boost-users] [Accumulators] Questions about the library
by dhruva 10 Aug '09

10 Aug '09

Hello, > Date: Thu, 25 Jun 2009 03:41:49 -0700 (PDT) > From: adish > Subject: Re: [Boost-users] [Accumulators] Questions about the library > To: boost-users(a)lists.boost.org > > > > > > ? ?2. If my platform does not support HW floating point ops, and all my > > > ? ?sample values are, e.g. ints, is there a way to specify that the > > > *computation > > > ? ?*and *return value itself* (i.e. result_type, e.g. the mean) will be done > > > ? ?in ints (naturally, at the cost of precision)? > > > > I have used UDT as I needed some extra information from the results (as I was > working on data from a time series). The reverse, using a POD like int must be > possible. Instead of the first argument being float/double, use an int. If that > does not work, you can wrap an int in a UDT, implement some of the operators > like '=', '<', '/' depending on the accumulator you plan to use (I did that as I > needed to send a UDT). > > Using an int still returns a float -- which is quite sensible. You might be able to write a custom extractor which can return an int. I have not explored it. I always thought the type you give as input is the type that gets returned. > By UDT do you mean User Defined Type? > This is an interesting approach. > I didn't realize that the result_type depends on the value type in > that way, I didn't see any UDTs mentioned in the docs. Yes, I mean writing a class/struct. I had to overload '<', '>', '+=' and '/' and I use it for variance, mean, median, min, max and my implementation of percentile. Remember to specialize std::numeric_limits for the UDT you write. Else, the min and/or max will have values that you have not added in the accumulator (garbage values). I had a tough time debugging it and finding the solution. Also, on Windows, you need to derive your specialization from _Num_base. As for implementing a generic rolling type: 1. You need to store your values in a circular buffer. 2. Each time you add a value into the buffer, pass the contents to a pristine accumulator. This may be sub optimal but works. Especially for those calculations where it is not easy to remove the effect of the first data that you want to pop. I had to do it this way because I was dealing with time series. If I want a rolling computation for a 1 hour window, I might add 1 sample and may have to remove more than 1 sample to ensure first-last is 1 hour window in the buffer. -dhruva Love Cricket? Check out live scores, photos, video highlights and more. Click here http://cricket.yahoo.com

3 2

[mpi] irecv / send problem
by Nick Collier 06 Aug '09

06 Aug '09

I running into an issue where an irecv followed by a send results in deadlock. A simple test case, class Item { private: friend class boost::serialization::access; template<class Archive> void serialize(Archive& ar, const unsigned int version) { ar & val; } public: int val; Item() : val(1) { } }; struct Receipt { boost::mpi::request request; std::vector<Item> items; }; int main(int argc, char **argv) { mpi::environment env(argc, argv); mpi::communicator world; Receipt receipt; vector<Item> msg(100000); int other = world.rank() == 0 ? 1 : 0; cout << world.rank() << " irecv from " << other << endl; receipt.request = world.irecv(other, 0, receipt.items); cout << world.rank() << " sending to " << other << endl; world.send(other, 0, msg); receipt.request.wait(); cout << "Done" << endl; } Run with mpirun -np 2, this never completes. It does complete with vector<Item> msg(10) however. Nick

5 7

Boost.test applicability
by Joel Falcou 01 Aug '09

01 Aug '09

I have a alrge number of numeric functions (~300) overloaded on various types. What I want to do is having some automated test that check if those functions returns proper values and behave correctly in term of precision. I get a look at Boost.Test but i can't really see if it's applicable to my need and, if it is, how exactly am I supposed to generate that much test and cover all cases (random value testing, singular value test, nan/inf handling etc...). My first idea was to use Boost.preprcoessor file iteration to quickly generate BOOST_TEST_CASE functions but I'm not sure this is a scalable practice. Other point, I saw that there is check in Boost.test which check for an arbitrary precision result. However it uses raw value (like 0.001) where having a check based on ULP or RMS maybe better. Is there anyway to integrate those test using existing test framework ? Thanks in advance

3 4

Boost Serialization: undefined reference to `__assert'
by Kesseli Pascal 31 Jul '09

31 Jul '09

Hello everyone Today, I just happened to download the latest version of cygwin, including gcc 3.4.4. While a simple hello world works perfectly fine, trying to compile this code using a statically linked libserialization.a results in dozens of linker errors, all stating "undefined reference to `__assert'". #include <boost/assign/list_of.hpp> #include <boost/archive/xml_iarchive.hpp> #include <boost/archive/xml_oarchive.hpp> #include <boost/lambda/lambda.hpp> #include <boost/serialization/vector.hpp> #include <iostream> #include <fstream> using namespace boost::lambda; using namespace boost::assign; using namespace boost::archive; using namespace boost::serialization; using namespace std; int main() { { vector<string> v = list_of("123")("Boost")("Serialization"); ofstream ofs("test.xml"); xml_oarchive oxml(ofs); oxml << make_nvp("VectorTest", v); } { vector<string> v; ifstream ifs("test.xml"); xml_iarchive ixml(ifs); ixml >> make_nvp("VectorTest", v); for_each(v.begin(), v.end(), cout << _1 << "\n"); } return 0; } Is this a problem or known issue when using cygwin? I'm using eclipse in addition, by the way. The code works perfectly fine on other platforms, so I imagine it must be a problem when using the serialization library with cygwin. Thanks in advance for any help you can offer me here and best regards Pascal Kesseli

4 3

UTF-16
by Robert Dailey 24 Jul '09

24 Jul '09

Hi everyone, I did a bit of googling to see if Boost 1.39 as any portable support for UTF-16 encoded strings, but I did not find any. I'm currently using wxWidgets in my application, and I need a decent string object to use. I know that wxWidgets has UTF-16 string support through wxString, however I do not want to expose this object in my interfaces. I want to remain as abstracted away from wxWidgets as possible. Having said that, if someone could tell me if there is any existing UTF-16 string support in Boost, I'd appreciate it. I did not find anything in the vault, sandbox, or trunk in Boost. If boost has no such string object, could someone give me a head start on where to look? Thanks.

9 15

[math] finding c such that w1/c+...+wn/c < math::tools::max_value<T>()
by er 21 Jul '09

21 Jul '09

Hi All, A little while back the following code was suggested to find c such that (x/c)+(y/c) <= C = math::tools::max_value<T>() template<class T>// static T scale_to_max( const T& x, const T& y ) { double max = (std::numeric_limits<T>::max)(); if(x > max - y) { double c = x/max + y/max; while(!boost::math::isinf(x/c + y/c)) { c = boost::math::float_prior(c); } while(boost::math::isinf(x/c + y/c)) { c = boost::math::float_next(c); } return(c); } else { return(static_cast<T>(1)); } } I'm now trying to generalize this i.e. find c such that w1/c+...+wn/c <= C: 1) split w1,..,wn into maximal size, sum less than C, subgroups. 2) apply scale_to_max iteratively over adjacent groups. This does not always suceed (see example at the bottom). Any suggestion would be appreciated. *.hpp template<typename T> class scale_sum_to_max{ public: typedef void result_type; typedef T argument_type; scale_sum_to_max(T cum_sum):cum_sum_(cum_sum),c_(static_cast<T>(1)){} scale_sum_to_max():cum_sum_(static_cast<T>(0)),c_(static_cast<T>(1)){} void operator()(T x){ if( math::isinf(scaled_sum()+(x/c_)) ){ T s = scale_to_max( scaled_sum(), x/c_ ); cum_sum_ /= s; c_ *= s; BOOST_ASSERT(!math::isinf(cum_sum_+x/c_)); } cum_sum_ += x/c_; } T scaling_factor()const{ return c_; } T scaled_sum()const{ return cum_sum_; } private: T cum_sum_; T c_; }; template<typename T> class finite_sums{ typedef std::vector<T> cont_t; public: typedef void result_type; typedef T argument_type; finite_sums() { coll_.push_back(static_cast<T>(0)); } void operator()(T x){ if(math::isinf(sum()+x)){ coll_.push_back(x); }else{ coll_.back() += x; } } const cont_t& collected()const{ return coll_; } T scaling_factor()const{ return for_each( begin(coll_), end(coll_), scale_sum_to_max<T>() ).scaling_factor(); } private: T sum()const{ return coll_.back(); } cont_t coll_; }; }} *.cpp typedef double value_t; typedef std::vector<value_t> vec_t; const unsigned n = 1e5; //n = 1e3 OK value_t w = math::tools::max_value<value_t>(); w /= 2; vec_t vec_w(n,w); value_t scaling_factor = for_each( begin(vec_w), end(vec_w), math_limit::finite_sums<value_t>() ).scaling_factor(); //assertion failed: (!math::isinf(cum_sum_+x/c_))

2 2

[Math Toolkit] Gamma Distribution
by Christian Henning 21 Jul '09

21 Jul '09

Hi there, the documention of the gamma distribution talks about two different forms. How can can generate the second one with shape=1 and scale=1? I know how to get the first form: boost::math::gamma_distribution<> dist( 1, 1 ); I don't see how to get the second one? Thanks, Christian

5 13

[Q] regular expressions change the 'substitue with' value
by V S P 14 Jul '09

14 Jul '09

Looking for some help with how to use regular expression library I have a string "!1!" I am replacing it with "\temp\i\otrq92rq40100__db1003.xls" ( a windows path to a file) This is how I am doing it std::string repl=subst_designator+ boost::lexical_cast<std::string>(counter)+ subst_designator; boost::algorithm::replace_all_regex(new_arg, boost::regex(repl),*res_it); Where: subst_designator = '!' counter = 1 new_arg = "!1!" *res_it="\temp\i\otrq92rq40100__db1003.xls" after the above code, the new_arg contains a strange string that appears to have back slashes removed and the first \t replaced with something like a 'tab' new_arg=" empiotrq92rq40100__db1003.xls" I am thinking that this is wrong, because the regular expression should copy without modifications the value of *res_it into the new_arg I am using 1.38 on windows with VS9. My question is , how can make it so the string in *res_it substitutes !1! thank you in advance, Vlad -- Vlad P author of C++ ORM http://github.com/vladp/CppOrm/tree/master -- http://www.fastmail.fm - IMAP accessible web-mail

3 7

[array] GCC 4.2 warnings with array initialization
by Allen Cronce 13 Jul '09

13 Jul '09

Hi all, When building with boost 1.36 and GCC 4.2 under Xcode, the following code produces a warning: boost::array<int, 2> warningArray = { 1, 2 }; warning: missing braces around initializer for 'int [2]' We see no warnings with GCG 4.0. We have a zero warning tolerance. All warnings are counted as errors. The boost::array documentation indicates that the single brace approach should work for conforming compilers, according to 8.5.1 (11) of the Standard. So does that mean that GCC 4.2 is non-conforming? Sure enough, when I change the code to include a second set of braces, the warning is gone: boost::array<int, 2> noWarningArray = { { 1, 2 } }; But this is a problem as we move to GCC 4.2 because a) we use initialized arrays all over the place, so this means a lot of little changes, and b) requiring an extra set of braces is counter intuitive and irritating at best. I don't like the idea of making the above code changes, and I don't want to globally disable the warning because it might be useful for other code. Does anyone have a better solution? Thanks in advance for any suggestions. Best, -- Allen Cronce

4 3