adding weight support for mean accumulator (#212)

changed type of sum to RealType for mean and added serialization tests for builtin accumulators

include/boost/histogram/accumulators/mean.hpp

@@ -28,12 +28,19 @@ public:
       : sum_(n), mean_(mean), sum_of_deltas_squared_(variance * (sum_ - 1)) {}
 
   void operator()(const RealType& x) {
-    sum_ += 1;
+    sum_ += static_cast<RealType>(1);
     const auto delta = x - mean_;
     mean_ += delta / sum_;
     sum_of_deltas_squared_ += delta * (x - mean_);
   }
 
+  void operator()(const RealType& w, const RealType& x) {
+    sum_ += w;
+    const auto delta = x - mean_;
+    mean_ += w * delta / sum_;
+    sum_of_deltas_squared_ += w * delta * (x - mean_);
+  }
+
   template <class T>
   mean& operator+=(const mean<T>& rhs) {
     const auto tmp = mean_ * sum_ + static_cast<RealType>(rhs.mean_ * rhs.sum_);
@@ -60,7 +67,7 @@ public:
     return !operator==(rhs);
   }
 
-  std::size_t count() const noexcept { return sum_; }
+  const RealType& count() const noexcept { return sum_; }
   const RealType& value() const noexcept { return mean_; }
   RealType variance() const { return sum_of_deltas_squared_ / (sum_ - 1); }
 
@@ -68,8 +75,7 @@ public:
   void serialize(Archive&, unsigned /* version */);
 
 private:
-  std::size_t sum_ = 0;
+  RealType sum_ = 0, mean_ = 0, sum_of_deltas_squared_ = 0;
-  RealType mean_ = 0, sum_of_deltas_squared_ = 0;
 };
 
 } // namespace accumulators

include/boost/histogram/serialization.hpp

@@ -32,6 +32,7 @@
 #include <boost/serialization/string.hpp>
 #include <boost/serialization/throw_exception.hpp>
 #include <boost/serialization/vector.hpp>
+#include <boost/serialization/version.hpp>
 #include <tuple>
 #include <type_traits>
 
@@ -44,6 +45,18 @@
 
 #ifndef BOOST_HISTOGRAM_DOXYGEN_INVOKED
 
+namespace boost {
+namespace serialization {
+// version 1 for boost::histogram::accumulators::mean<RealType>
+template <class RealType>
+struct version<::boost::histogram::accumulators::mean<RealType>> {
+  typedef mpl::int_<1> type;
+  typedef mpl::integral_c_tag tag;
+  BOOST_STATIC_CONSTANT(int, value = version::type::value);
+};
+} // namespace serialization
+} // namespace boost
+
 namespace std {
 template <class Archive, class... Ts>
 void serialize(Archive& ar, tuple<Ts...>& t, unsigned /* version */) {
@@ -72,8 +85,14 @@ void weighted_sum<RealType>::serialize(Archive& ar, unsigned /* version */) {
 
 template <class RealType>
 template <class Archive>
-void mean<RealType>::serialize(Archive& ar, unsigned /* version */) {
+void mean<RealType>::serialize(Archive& ar, unsigned version) {
-  ar& serialization::make_nvp("sum", sum_);
+  if (version == 0 && Archive::is_loading::value) {
+    std::size_t sum;
+    ar& serialization::make_nvp("sum", sum);
+    sum_ = static_cast<RealType>(sum);
+  } else {
+    ar& serialization::make_nvp("sum", sum_);
+  }
   ar& serialization::make_nvp("mean", mean_);
   ar& serialization::make_nvp("sum_of_deltas_squared", sum_of_deltas_squared_);
 }

test/CMakeLists.txt

@@ -8,6 +8,8 @@ boost_test(SOURCES check_cmake_version.cpp ARGUMENTS ${PROJECT_VERSION}
 # keep in sync with Jamfile, this should be automatized...
 boost_test(TYPE compile-fail SOURCES make_histogram_fail0.cpp)
 boost_test(TYPE compile-fail SOURCES make_histogram_fail1.cpp)
+boost_test(TYPE run SOURCES accumulators_test.cpp
+  LIBRARIES Boost::histogram Boost::core)
 boost_test(TYPE run SOURCES algorithm_project_test.cpp
   LIBRARIES Boost::histogram Boost::core)
 boost_test(TYPE run SOURCES algorithm_reduce_test.cpp
@@ -78,8 +80,6 @@ boost_test(TYPE run SOURCES histogram_test.cpp
   LIBRARIES Boost::histogram Boost::core)
 boost_test(TYPE run SOURCES indexed_test.cpp
   LIBRARIES Boost::histogram Boost::core)
-boost_test(TYPE run SOURCES internal_accumulators_test.cpp
-  LIBRARIES Boost::histogram Boost::core)
 boost_test(TYPE run SOURCES storage_adaptor_test.cpp
   LIBRARIES Boost::histogram Boost::core)
 boost_test(TYPE run SOURCES unlimited_storage_test.cpp

test/Jamfile

@@ -22,6 +22,7 @@ project
     ;
 
 alias cxx14 :
+    [ run accumulators_test.cpp ]
     [ run algorithm_project_test.cpp ]
     [ run algorithm_reduce_test.cpp ]
     [ run algorithm_sum_test.cpp ]
@@ -57,7 +58,6 @@ alias cxx14 :
     [ run histogram_operators_test.cpp ]
     [ run histogram_test.cpp ]
     [ run indexed_test.cpp ]
-    [ run internal_accumulators_test.cpp ]
     [ run storage_adaptor_test.cpp ]
     [ run unlimited_storage_test.cpp ]
     [ run utility_test.cpp ]
@@ -86,6 +86,7 @@ alias accumulators : [ run boost_accumulators_support_test.cpp ] : <warnings>off
 alias range : [ run boost_range_support_test.cpp ] : <warnings>off ;
 alias units : [ run boost_units_support_test.cpp ] : <warnings>off ;
 alias serialization :
+    [ run accumulators_serialization_test.cpp libserial : $(THIS_PATH) ]
     [ run axis_variant_serialization_test.cpp libserial : $(THIS_PATH) ]
     [ run histogram_serialization_test.cpp libserial : $(THIS_PATH) ]
     [ run storage_adaptor_serialization_test.cpp libserial : $(THIS_PATH) ]

test/accumulators_serialization_test.cpp

@@ -0,0 +1,90 @@
+// Copyright 2019 Hans Dembinski
+//
+// Distributed under the Boost Software License, Version 1.0.
+// (See accompanying file LICENSE_1_0.txt
+// or copy at http://www.boost.org/LICENSE_1_0.txt)
+
+#include <boost/assert.hpp>
+#include <boost/core/lightweight_test.hpp>
+#include <boost/histogram/accumulators.hpp>
+#include <boost/histogram/serialization.hpp>
+#include "throw_exception.hpp"
+#include "utility_serialization.hpp"
+
+using namespace boost::histogram;
+
+int main(int argc, char** argv) {
+  BOOST_ASSERT(argc == 2);
+
+  // mean v0
+  {
+    const auto filename = join(argv[1], "accumulators_serialization_test_mean_v0.xml");
+    accumulators::mean<> a;
+    load_xml(filename, a);
+    BOOST_TEST_EQ(a.count(), 3);
+    BOOST_TEST_EQ(a.value(), 2);
+    BOOST_TEST_EQ(a.variance(), 0.5);
+  }
+
+  // mean
+  {
+    const auto filename = join(argv[1], "accumulators_serialization_test_mean.xml");
+    accumulators::mean<> a;
+    a(1);
+    a(0.5, 2);
+    a(3);
+    print_xml(filename, a);
+
+    accumulators::mean<> b;
+    BOOST_TEST_NOT(a == b);
+    load_xml(filename, b);
+    BOOST_TEST(a == b);
+  }
+
+  // sum
+  {
+    const auto filename = join(argv[1], "accumulators_serialization_test_sum.xml");
+    accumulators::sum<> a;
+    a += 1e100;
+    a += 1;
+    print_xml(filename, a);
+
+    accumulators::sum<> b;
+    BOOST_TEST_NOT(a == b);
+    load_xml(filename, b);
+    BOOST_TEST(a == b);
+  }
+
+  // weighted_mean
+  {
+    const auto filename =
+        join(argv[1], "accumulators_serialization_test_weighted_mean.xml");
+    accumulators::weighted_mean<> a;
+    a(1);
+    a(0.5, 2);
+    a(3);
+    print_xml(filename, a);
+
+    accumulators::weighted_mean<> b;
+    BOOST_TEST_NOT(a == b);
+    load_xml(filename, b);
+    BOOST_TEST(a == b);
+  }
+
+  // weighted_sum
+  {
+    const auto filename =
+        join(argv[1], "accumulators_serialization_test_weighted_sum.xml");
+    accumulators::weighted_sum<> a;
+    a += 1;
+    a += 10;
+    print_xml(filename, a);
+
+    accumulators::weighted_sum<> b;
+    BOOST_TEST_NOT(a == b);
+    load_xml(filename, b);
+    BOOST_TEST(a == b);
+  }
+
+  return boost::report_errors();
+}

test/accumulators_serialization_test_mean.xml

@@ -0,0 +1,17 @@
+<!--
+	Copyright 2019 Hans Dembinski
+
+	Distributed under the Boost Software License, Version 1.0.
+	(See accompanying file LICENSE_1_0.txt
+	or copy at http://www.boost.org/LICENSE_1_0.txt)
+-->
+
+<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+<!DOCTYPE boost_serialization>
+<boost_serialization signature="serialization::archive" version="17">
+<item class_id="0" tracking_level="0" version="1">
+	<sum>2.50000000000000000e+00</sum>
+	<mean>2.00000000000000000e+00</mean>
+	<sum_of_deltas_squared>2.00000000000000000e+00</sum_of_deltas_squared>
+</item>
+</boost_serialization>

test/accumulators_serialization_test_mean_v0.xml

@@ -0,0 +1,17 @@
+<!--
+	Copyright 2019 Hans Dembinski
+
+	Distributed under the Boost Software License, Version 1.0.
+	(See accompanying file LICENSE_1_0.txt
+	or copy at http://www.boost.org/LICENSE_1_0.txt)
+-->
+
+<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+<!DOCTYPE boost_serialization>
+<boost_serialization signature="serialization::archive" version="17">
+<item class_id="0" tracking_level="0" version="0">
+	<sum>3</sum>
+	<mean>2.00000000000000000e+00</mean>
+	<sum_of_deltas_squared>1.00000000000000000e+00</sum_of_deltas_squared>
+</item>
+</boost_serialization>

test/accumulators_serialization_test_sum.xml

@@ -0,0 +1,16 @@
+<!--
+	Copyright 2019 Hans Dembinski
+
+	Distributed under the Boost Software License, Version 1.0.
+	(See accompanying file LICENSE_1_0.txt
+	or copy at http://www.boost.org/LICENSE_1_0.txt)
+-->
+
+<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+<!DOCTYPE boost_serialization>
+<boost_serialization signature="serialization::archive" version="17">
+<item class_id="0" tracking_level="0" version="0">
+	<large>1.00000000000000002e+100</large>
+	<small>1.00000000000000000e+00</small>
+</item>
+</boost_serialization>

test/accumulators_serialization_test_weighted_mean.xml

@@ -0,0 +1,18 @@
+<!--
+	Copyright 2019 Hans Dembinski
+
+	Distributed under the Boost Software License, Version 1.0.
+	(See accompanying file LICENSE_1_0.txt
+	or copy at http://www.boost.org/LICENSE_1_0.txt)
+-->
+
+<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+<!DOCTYPE boost_serialization>
+<boost_serialization signature="serialization::archive" version="17">
+<item class_id="0" tracking_level="0" version="0">
+	<sum_of_weights>2.50000000000000000e+00</sum_of_weights>
+	<sum_of_weights_squared>2.25000000000000000e+00</sum_of_weights_squared>
+	<weighted_mean>2.00000000000000000e+00</weighted_mean>
+	<sum_of_weighted_deltas_squared>2.00000000000000000e+00</sum_of_weighted_deltas_squared>
+</item>
+</boost_serialization>

test/accumulators_serialization_test_weighted_sum.xml

@@ -0,0 +1,16 @@
+<!--
+	Copyright 2019 Hans Dembinski
+
+	Distributed under the Boost Software License, Version 1.0.
+	(See accompanying file LICENSE_1_0.txt
+	or copy at http://www.boost.org/LICENSE_1_0.txt)
+-->
+
+<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>
+<!DOCTYPE boost_serialization>
+<boost_serialization signature="serialization::archive" version="17">
+<item class_id="0" tracking_level="0" version="0">
+	<sum_of_weights>1.10000000000000000e+01</sum_of_weights>
+	<sum_of_weights_squared>1.01000000000000000e+02</sum_of_weights_squared>
+</item>
+</boost_serialization>

test/accumulators_test.cpp

@@ -11,9 +11,9 @@
 #include <boost/histogram/accumulators/thread_safe.hpp>
 #include <boost/histogram/accumulators/weighted_mean.hpp>
 #include <boost/histogram/accumulators/weighted_sum.hpp>
-#include "throw_exception.hpp"
 #include <sstream>
 #include "is_close.hpp"
+#include "throw_exception.hpp"
 
 using namespace boost::histogram;
 using namespace std::literals;
@@ -98,6 +98,15 @@ int main() {
     BOOST_TEST_EQ(c.count(), 8);
     BOOST_TEST_EQ(c.value(), 10);
     BOOST_TEST_IS_CLOSE(c.variance(), 25.714, 1e-3);
+
+    // also same as feeding all samples twice
+    m_t d;
+    d(2, 4);
+    d(2, 7);
+    d(2, 13);
+    d(2, 16);
+
+    BOOST_TEST_EQ(d, c);
   }
 
   {

test/histogram_fill_test.cpp

@@ -241,7 +241,7 @@ void run_tests() {
     BOOST_TEST_EQ(h, h2);
   }
 
-  // 1D with samples
+  // 1D profile with samples
   {
     auto h = make_s(Tag(), profile_storage(), in(1, 3));
     auto h2 = h;
@@ -253,9 +253,14 @@ void run_tests() {
     h2.fill(x, sample(x));
 
     BOOST_TEST_EQ(h, h2);
+
+    for (auto&& xi : x) h(xi, sample(xi), weight(xi));
+    h2.fill(x, sample(x), weight(x));
+
+    BOOST_TEST_EQ(h, h2);
   }
 
-  // 2D with samples and weights
+  // 2D weighted profile with samples and weights
   {
     auto h = make_s(Tag(), weighted_profile_storage(), in(1, 3), in0(1, 3));
     auto h2 = h;