Experimental prime sieves

.gitignore

@@ -7,3 +7,6 @@ inspect.html
 test/cuda
 *.DS_Store
 /**/*.dSYM/
+build/*
+*.json
+

doc/math.qbk

@@ -1,13 +1,13 @@
 [book Math Toolkit
     [quickbook 1.7]
-    [copyright 2006-2020 Nikhar Agrawal, Anton Bikineev, Paul A. Bristow, Marco Guazzone, Christopher Kormanyos, Hubert Holin, Bruno Lalande, John Maddock, Evan Miller, Jeremy Murphy, Matthew Pulver, Johan Råde, Gautam Sewani, Benjamin Sobotta, Nicholas Thompson, Thijs van den Berg, Daryle Walker and Xiaogang Zhang]
+    [copyright 2006-2020 Nikhar Agrawal, Anton Bikineev, Matthew Borland, Paul A. Bristow, Marco Guazzone, Christopher Kormanyos, Hubert Holin, Bruno Lalande, John Maddock, Evan Miller, Jeremy Murphy, Matthew Pulver, Johan Råde, Gautam Sewani, Benjamin Sobotta, Nicholas Thompson, Thijs van den Berg, Daryle Walker and Xiaogang Zhang]
     [/purpose ISBN 0-9504833-2-X  978-0-9504833-2-0, Classification 519.2-dc22]
     [license
         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])
     ]
-    [authors [Agrawal, Nikhar], [Bikineev, Anton], [Bristow, Paul A.], [Holin, Hubert], [Guazzone, Marco], [Kormanyos, Christopher], [Lalande, Bruno], [Maddock, John], [Miller, Evan], [Murphy, Jeremy W.], [Pulver, Matthew], [Råde, Johan], [Sobotta, Benjamin], [Sewani, Gautam], [Thompson, Nicholas], [van den Berg, Thijs], [Walker, Daryle], [Zhang, Xiaogang]]
+    [authors [Agrawal, Nikhar], [Bikineev, Anton], [Borland, Matthew], [Bristow, Paul A.], [Holin, Hubert], [Guazzone, Marco], [Kormanyos, Christopher], [Lalande, Bruno], [Maddock, John], [Miller, Evan], [Murphy, Jeremy W.], [Pulver, Matthew], [Råde, Johan], [Sobotta, Benjamin], [Sewani, Gautam], [Thompson, Nicholas], [van den Berg, Thijs], [Walker, Daryle], [Zhang, Xiaogang]]
     [/last-revision $Date$]
     [version 2.12.0]
 ]

doc/sf/number_series.qbk

@@ -259,9 +259,86 @@ Passing a value greater than `max_prime` results in a __domain_error being raise
 
 This function is `constexpr` only if the compiler supports C++14 constexpr functions.
 
-[endsect] [/section:primes]
+[endsect] [/section:primes Prime Numbers]
 
-[endsect] [/Number Series]
+[section:prime_sieve Prime Sieve]
+
+[h4 Synopsis]
+
+``
+#include <boost/math/special_functions/prime_sieve.hpp>
+``
+
+namespace boost { namespace math {
+
+    template<class ExecutionPolicy, typename Integer, class Container>
+    void prime_sieve(ExecutionPolicy&& policy, Integer upper_bound, Container &resultant_primes)
+
+    template<class ExecutionPolicy, class Integer, class Container>
+    void prime_range(ExecutionPolicy&& policy, Integer lower_bound, Integer upper_bound, Container &resultant_primes)
+
+    template<class Integer, class Container>
+    void prime_sieve(Integer upper_bound, Container &resultant_primes)
+
+    template<class Integer, class Container>
+    void prime_range(Integer lower_bound, Integer upper_bound, Container &resultant_primes)
+
+    template<class Integer>
+    constexpr void prime_reserve(Integer upper_bound, std::vector<Integer> &prime_container)
+
+
+}} // End namespaces
+
+[h4 Description]
+
+There are two sets of prime sieveing functions available: `prime_sieve` and `prime_range`. 
+`prime_sieve` will return all primes in the range [2, `upper_bound`). 
+`prime_range` will return all primes in the range [lower_bound, upper_bound).
+Both `prime_sieve` and `prime_range` can perform arbitrary precision calculations using `boost::multiprecision::cpp_int` or `boost::multiprecision::mpz_int`. 
+
+`prime_sieve` and `prime_range` both have a parameter for an execution policy.
+The policies `std::execution::par` or `std::execution::par_unseq` will enable internal multi-threading.
+All other policies or no policy will result in the `prime_sieve` and `prime_range` executing sequentially. 
+
+`prime_reserve` uses the prime number theorem to reserve for approximately the correct number of primes given `upper_bound`.
+
+[h4 Examples]
+    // To reserve space and calculate primes [2, 1,000,000) in parallel
+    std::vector<int64_t> primes;
+    boost::math::prime_reserve(1'000'000, primes);
+    boost::math::prime_sieve(std::execution::par, 1'000'000, primes);
+
+    // To calculate primes [100, 1,000) sequentially
+    std::vector<int64_t> primes;
+    boost::math::prime_range(100, 1'000, primes);
+
+[h4 Complexity]
+These functions were tested for complexity using [@https://github.com/google/benchmark google benchmark] on a system with the following specifications:
+* CPU: Intel i5-10500
+* OS: Red Hat Enterprise Linux 8.2
+* Compiler: gcc-g++ 10.2.0 (Compiled from source with GMP 6.2.0, MPC 1.2.0, and MPFR 4.1.0 using RHEL gcc-toolset-9)
+* Compiler flags: -Ofast -MMD -march=native -g -lbenchmark -lpthread -lgmp
+
+Range of test upper_bound: 2[super 1] - 2[super 30]
+
+[pre'''
+[table:id Complexity Calculations `bigo[](N)`
+  [[Type]     [Time Complexity]   [Time Complexity RMS]   [CPU Complexity]    [CPU Complexity RMS]]
+  [[int32_t]  [0.72N]             [5%]                    [0.02N]             [13%]]
+  [[int64_t]  [0.78N]             [16%]                   [0.03N]             [6%]]
+  [[uint32_t] [0.71N]             [10%]                   [0.02N]             [14%]]
+  [[cpp_int]  [6.24N]             [44%]                   [0.35N]             [4%]]
+  [[mpz_int]  [9.27N]             [11%]                   [0.69N]             [8%]]
+]
+''']
+
+[h4 References]
+* Sorensen, Jonathan [@https://research.cs.wisc.edu/techreports/1990/TR909.pdf An Introduction to Prime Number Sieves]
+* Gries, David and Misra, Jayadev [@https://www.cs.utexas.edu/users/misra/scannedPdf.dir/linearSieve.pdf A Linear Sieve Algorithm for Finding Prime Numbers]
+
+[endsect] [/section:primes Prime Sieve]
+
+[endsect] [/section:number_series Number Series]
 
 [/
   Copyright 2013, 2014 Nikhar Agrawal, Christopher Kormanyos, John Maddock, Paul A. Bristow.