C++20 includes updated versions of the many algorithms in the <algorithm> header. This header supplements these with updated versions of other algorithms from the <numeric> header.

If you're using C++20 ranges (or NanoRange -- see below) then you can drop the numeric_ranges.hpp header into your project and use it as a modern replacement for <numeric> until such time as C++23 comes along.

constexpr std::array arr{1, 2, 3, 4};
std::vector<int> out;
tcb::partial_sum(arr, std::back_inserter(out));
// out contains [1, 3, 6, 10]
const int prod = tcb::inner_product(arr, out, 0);
// prod = (1 * 1) + (2 * 3) + (3 * 6) + (4 * 10) = 65
assert(prod == 65);
constexpr auto sq = [](int i) { return i * i; };
constexpr int sum = tcb::accumulate(arr, 0, {}, sq);
// sum = 1 + 4 + 9 + 16 = 30
static_assert(sum == 30);

If your standard library provides an implementation of C++20 ranges, you can just copy numeric_ranges.hpp into your project and use it as an alternative to the <numeric> header.

The rest of this respository contains testing machinery and is not required for use.

numeric_ranges.hpp is also compatible with NanoRange. To use it, define the symbol TCB_NUMERIC_RANGES_USE_NANORANGE before #include-ing this header, for example

#define TCB_NUMERIC_RANGES_USE_NANORANGE
#include <numeric_ranges.hpp>

(or by using an equivalent compiler define).

The following algorithms are provided in this header:

• iota
• accumulate / reduce
• inner_product / transform_reduce
• adjacent_difference
• partial_sum

Note that in this implementation, reduce and transform_reduce always perform their operations in order and so are equivalent to accumulate and inner_product respectively.

Unlike the other algorithms in std::ranges (and the eventual C++23 version of <numeric>) the implementations in this library are unconstrained.

This means that it's possible to get horrible, C++98-style template error messages when things go wrong. It's also possible that in rare cases, code you write today with this library may fail constraint checks when they are added in future.

This library is provided under the terms of the Boost licence.

Thanks to Eric Niebler for writing the Range-V3 tests for all these algorithms, which I have lovingly "borrowed".

Thanks to Phil Nash et al for the fantastic Catch2 testing framework.