histogram/doc/getting_started.qbk

[section:getting_started Getting started]

To get you started quickly, here are some heavily commented examples to copy paste from. If you prefer a more traditional, structured exposition, check out the [link histogram.guide full user guide].

[section Make and use a static 1d-histogram in C++]

If possible, use the static histogram. It is faster and user errors are caught at compile time.

[c++]``
#include <boost/histogram.hpp>
#include <iostream>

int main(int, char**) {
    namespace bh = boost::histogram;
    using namespace bh::literals; // enables _c suffix

    /*
        create a static 1d-histogram with an axis that has 10 equidistant
        bins on the real line from -1.0 to 2.0, and label it as "x"
    */
    auto h = bh::make_static_histogram(
        bh::axis::regular<>(10, -1.0, 2.0, "x")
    );

    // fill histogram with data, typically this would happen in a loop
    h.fill(-1.5); // put in underflow bin
    h.fill(-1.0); // included in first bin, bin interval is semi-open
    h.fill(-0.5);
    h.fill(1.1);
    h.fill(0.3);
    h.fill(1.7);
    h.fill(2.0);  // put in overflow bin, bin interval is semi-open
    h.fill(20.0); // put in overflow bin

    /*
        use bh::count(N) if you would otherwise call h.fill(...) with
        *same* argument N times, N is an integer argument
    */
    h.fill(1.0, bh::count(4));

    /*
        do a weighted fill using bh::weight, which accepts a double
        - don't mix this with bh::count, both have a different effect on the
          variance (see Rationale for an explanation regarding weights)
        - if you don't know what this is good for, use bh::count instead,
          it is most likeliy what you want and it is more efficient
    */
    h.fill(0.1, bh::weight(2.5));

    /*
        iterate over bins, loop excludes under- and overflow bins
        - index 0_c is a compile-time number, the only way in C++ to make
          axis(...) to return a different type for each index
        - for-loop yields instances of `std::pair<int, bin_type>`, where
          `bin_type` usually is a semi-open interval representing the bin,
          whose edges can be accessed with methods `lower()` and `upper()`,
          but the [bin type] depends on the axis, look it up in the reference
        - `value(index)` method returns the bin count at index
        - `variance(index)` method returns a variance estimate of the bin
          count at index (see Rationale section for what this means)
    */
    for (const auto& bin : h.axis(0_c)) {
        std::cout << "bin " << bin.first << " x in ["
                  << bin.second.lower() << ", " << bin.second.upper() << "): "
                  << h.value(bin.first) << " +/- "
                  << std::sqrt(h.variance(bin.first))
                  << std::endl;
    }

    // accessing under- and overflow bins is easy, use indices -1 and 10
    std::cout << "underflow bin [" << h.axis(0_c)[-1].lower()
              << ", " << h.axis(0_c)[-1].upper() << "): "
              << h.value(-1) << " +/- " << std::sqrt(h.variance(-1))
              << std::endl;
    std::cout << "overflow  bin [" << h.axis(0_c)[10].lower()
              << ", " << h.axis(0_c)[10].upper() << "): "
              << h.value(10) << " +/- " << std::sqrt(h.variance(10))
              << std::endl;

    /* program output:

    bin 0 x in [-1, -0.7): 1 +/- 1
    bin 1 x in [-0.7, -0.4): 1 +/- 1
    bin 2 x in [-0.4, -0.1): 0 +/- 0
    bin 3 x in [-0.1, 0.2): 2.5 +/- 2.5
    bin 4 x in [0.2, 0.5): 1 +/- 1
    bin 5 x in [0.5, 0.8): 0 +/- 0
    bin 6 x in [0.8, 1.1): 4 +/- 2
    bin 7 x in [1.1, 1.4): 1 +/- 1
    bin 8 x in [1.4, 1.7): 0 +/- 0
    bin 9 x in [1.7, 2): 1 +/- 1
    underflow bin [-inf, -1): 1 +/- 1
    overflow  bin [2, inf): 2 +/- 1.41421

    */
}
``

[endsect]

[section Make and use a dynamic 3d-histogram in C++]

Dynamic histograms are a bit slower than static histograms, but still faster than other libraries. Use a dynamic histogram when you only know at runtime how to layout the histogram, for example, because you wrote a graphical user interface that uses Boost.Histogram underneath.

[c++]``
#include <boost/histogram.hpp>
#include <boost/random/mersenne_twister.hpp>
#include <boost/random/normal_distribution.hpp>
#include <cstdlib>
#include <string>

namespace br = boost::random;
namespace bh = boost::histogram;

int main() {
    /*
        create a dynamic histogram with the factory `make_dynamic_histogram`
        - axis can be passed directly just like for `make_static_histogram`
        - in addition, the factory also accepts iterators over a sequence of
          axis::any, the polymorphic type that can hold concrete axis types
    */
    std::vector<bh::axis::any<>> axes;
    axes.emplace_back(bh::axis::category<std::string>({"red", "blue"}));
    axes.emplace_back(bh::axis::regular<>(5, -5, 5, "x"));
    axes.emplace_back(bh::axis::regular<>(5, -5, 5, "y"));
    auto h = bh::make_dynamic_histogram(axes.begin(), axes.end());

    // fill histogram with random numbers
    br::mt19937 gen;
    br::normal_distribution<> norm;
    for (int i = 0; i < 1000; ++i)
        h.fill(i % 2 ? "red" : "blue", norm(gen), norm(gen));

    /*
        print dynamic histogram by iterating over bins
        - for most axis types, the for loop looks just like for a static
          histogram, except that we can pass runtime numbers, too
        - if the [bin type] of the axis is not convertible to a
          double interval, one needs to cast axis::any before looping;
          this is here the case for the category axis
    */
    using cas = bh::axis::category<std::string>;
    for (const auto& cbin : bh::axis::cast<cas>(h.axis(0))) {
        std::printf("%s\n", cbin.second.c_str());
        for (const auto& ybin : h.axis(2)) { // rows
            for (const auto& xbin : h.axis(1)) { // columns
                std::printf("%3.0f ", h.value(cbin.first, xbin.first, ybin.first));
            }
            std::printf("\n");
        }
    }
}
``
[note
If you care about maximum performance: In this example, `axis::category<std::string>` is used with two string labels "red" and "blue". It is faster to use an enum, `enum { red, blue };` and a `axis::category<>` axis.
]

[endsect]

[section Make and use a 2d-histogram in Python]

You need to build the library with Numpy support to run this example.

[python]``
import histogram as hg
import numpy as np

# create 2d-histogram with two axes with 10 equidistant bins from -3 to 3
h = hg.histogram(hg.axis.regular(10, -3, 3, "x"),
                 hg.axis.regular(10, -3, 3, "y"))

# generate some numpy arrays with data to fill into histogram,
# in this case normal distributed random numbers in x and y
x = np.random.randn(1000)
y = 0.5 * np.random.randn(1000)

# fill histogram with numpy arrays, this is very fast
h.fill(x, y)

# get representations of the bin edges as Numpy arrays, this representation
# differs from `list(h.axis(0))`, because it is optimised for compatibility
# with existing Numpy code, i.e. to replace numpy.histogram
x = np.array(h.axis(0))
y = np.array(h.axis(1))

# creates a view of the counts (no copy involved)
count_matrix = np.asarray(h)

# cut off the under- and overflow bins to not confuse matplotib (no copy)
reduced_count_matrix = count_matrix[:-2,:-2]

try:
    # draw the count matrix
    import matplotlib.pyplot as plt
    plt.pcolor(x, y, reduced_count_matrix.T)
    plt.xlabel(h.axis(0).label)
    plt.ylabel(h.axis(1).label)
    plt.savefig("example_2d_python.png")
except ImportError:
    # ok, no matplotlib, then just print the full count matrix
    print count_matrix

    # output of the print looks something like this, the two right-most rows
    # and two down-most columns represent under-/overflow bins
    # [[ 0  0  0  1  5  0  0  1  0  0  0  0]
    #  [ 0  0  0  1 17 11  6  0  0  0  0  0]
    #  [ 0  0  0  5 31 26  4  1  0  0  0  0]
    #  [ 0  0  3 20 59 62 26  4  0  0  0  0]
    #  [ 0  0  1 26 96 89 16  1  0  0  0  0]
    #  [ 0  0  4 21 86 84 20  1  0  0  0  0]
    #  [ 0  0  1 24 71 50 15  2  0  0  0  0]
    #  [ 0  0  0  6 26 37  7  0  0  0  0  0]
    #  [ 0  0  0  0 11 10  2  0  0  0  0  0]
    #  [ 0  0  0  1  2  3  1  0  0  0  0  0]
    #  [ 0  0  0  0  0  2  0  0  0  0  0  0]
    #  [ 0  0  0  0  0  1  0  0  0  0  0  0]]
``

[endsect]

[section Make and use a 1d-histogram in Python without Numpy]

Building the library with Numpy support is highly recommended, but here is an example on how to use the library without Numpy support for completeness.

[python]``
import histogram as hg

# make 1-d histogram with 5 logarithmic bins from 1e0 to 1e5
h = hg.histogram(hg.axis.regular_log(5, 1e0, 1e5, "x"))

# fill histogram with numbers
for x in (2e0, 2e1, 2e2, 2e3, 2e4):
    h.fill(x, count=2)

# iterate over bins and access bin counter
for idx, (lower, upper) in enumerate(h.axis(0)):
    print "bin {0} x in [{1}, {2}): {3} +/- {4}".format(
        idx, lower, upper, h.value(idx), h.variance(idx) ** 0.5)
``

[endsect]

[endsect]