Implement std::cauchy_distribution

libcudacxx/include/cuda/std/__random/cauchy_distribution.h

@@ -0,0 +1,195 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+// SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _CUDA_STD___RANDOM_CAUCHY_DISTRIBUTION_H
+#define _CUDA_STD___RANDOM_CAUCHY_DISTRIBUTION_H
+
+#include <cuda/std/detail/__config>
+
+#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
+#  pragma GCC system_header
+#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
+#  pragma clang system_header
+#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
+#  pragma system_header
+#endif // no system header
+
+#include <cuda/std/__random/is_valid.h>
+#include <cuda/std/__random/uniform_real_distribution.h>
+#include <cuda/std/cmath>
+#include <cuda/std/limits>
+
+#include <cuda/std/__cccl/prologue.h>
+
+_CCCL_BEGIN_NAMESPACE_CUDA_STD
+
+template <class _RealType = double>
+class cauchy_distribution
+{
+  static_assert(__libcpp_random_is_valid_realtype<_RealType>, "RealType must be a supported floating-point type");
+
+public:
+  // types
+  using result_type = _RealType;
+
+  class param_type
+  {
+    result_type __a_ = result_type{0};
+    result_type __b_ = result_type{1};
+
+  public:
+    using distribution_type = cauchy_distribution;
+
+    _CCCL_API explicit param_type(result_type __a = 0, result_type __b = 1)
+        : __a_{__a}
+        , __b_{__b}
+    {}
+
+    [[nodiscard]] _CCCL_API constexpr result_type a() const noexcept
+    {
+      return __a_;
+    }
+    [[nodiscard]] _CCCL_API constexpr result_type b() const noexcept
+    {
+      return __b_;
+    }
+
+    [[nodiscard]] friend _CCCL_API constexpr bool operator==(const param_type& __x, const param_type& __y) noexcept
+    {
+      return __x.__a_ == __y.__a_ && __x.__b_ == __y.__b_;
+    }
+#if _CCCL_STD_VER <= 2017
+    [[nodiscard]] friend _CCCL_API constexpr bool operator!=(const param_type& __x, const param_type& __y) noexcept
+    {
+      return !(__x == __y);
+    }
+#endif // _CCCL_STD_VER <= 2017
+  };
+
+private:
+  param_type __p_;
+
+public:
+  // constructor and reset functions
+  _CCCL_API cauchy_distribution() noexcept
+      : cauchy_distribution{0}
+  {}
+  _CCCL_API explicit cauchy_distribution(result_type __a, result_type __b = 1) noexcept
+      : __p_{param_type{__a, __b}}
+  {}
+  _CCCL_API explicit cauchy_distribution(const param_type& __p) noexcept
+      : __p_{__p}
+  {}
+  _CCCL_API constexpr void reset() noexcept {}
+
+  // generating functions
+  template <class _URng>
+  [[nodiscard]] _CCCL_API result_type operator()(_URng& __g)
+  {
+    return (*this)(__g, __p_);
+  }
+  template <class _URng>
+  [[nodiscard]] _CCCL_API result_type operator()(_URng& __g, const param_type& __p)
+  {
+    static_assert(__cccl_random_is_valid_urng<_URng>, "URng must meet the UniformRandomBitGenerator requirements");
+    uniform_real_distribution<result_type> __gen;
+    // purposefully let tan arg get as close to pi/2 as it wants, tan will return a finite
+    return __p.a() + __p.b() * ::cuda::std::tan(3.1415926535897932384626433832795 * __gen(__g));
+  }
+
+  // property functions
+  [[nodiscard]] _CCCL_API constexpr result_type a() const noexcept
+  {
+    return __p_.a();
+  }
+  [[nodiscard]] _CCCL_API constexpr result_type b() const noexcept
+  {
+    return __p_.b();
+  }
+
+  [[nodiscard]] _CCCL_API constexpr param_type param() const noexcept
+  {
+    return __p_;
+  }
+  _CCCL_API constexpr void param(const param_type& __p) noexcept
+  {
+    __p_ = __p;
+  }
+
+  [[nodiscard]] _CCCL_API static constexpr result_type min() noexcept
+  {
+    return -numeric_limits<result_type>::infinity();
+  }
+  [[nodiscard]] _CCCL_API static constexpr result_type max() noexcept
+  {
+    return numeric_limits<result_type>::infinity();
+  }
+
+  [[nodiscard]] friend _CCCL_API constexpr bool
+  operator==(const cauchy_distribution& __x, const cauchy_distribution& __y) noexcept
+  {
+    return __x.__p_ == __y.__p_;
+  }
+#if _CCCL_STD_VER <= 2017
+  [[nodiscard]] friend _CCCL_API constexpr bool
+  operator!=(const cauchy_distribution& __x, const cauchy_distribution& __y) noexcept
+  {
+    return !(__x == __y);
+  }
+#endif // _CCCL_STD_VER <= 2017
+
+#if !_CCCL_COMPILER(NVRTC)
+  template <class _CharT, class _Traits>
+  friend ::std::basic_ostream<_CharT, _Traits>&
+  operator<<(::std::basic_ostream<_CharT, _Traits>& __os, const cauchy_distribution& __x)
+  {
+    using ostream_type                        = ::std::basic_ostream<_CharT, _Traits>;
+    using ios_base                            = typename ostream_type::ios_base;
+    const typename ios_base::fmtflags __flags = __os.flags();
+    const _CharT __fill                       = __os.fill();
+    const ::std::streamsize __precision       = __os.precision();
+    __os.flags(ios_base::dec | ios_base::left | ios_base::scientific);
+    _CharT __sp = __os.widen(' ');
+    __os.fill(__sp);
+    __os.precision(numeric_limits<result_type>::max_digits10);
+    __os << __x.a() << __sp << __x.b();
+    __os.flags(__flags);
+    __os.fill(__fill);
+    __os.precision(__precision);
+    return __os;
+  }
+
+  template <class _CharT, class _Traits>
+  friend ::std::basic_istream<_CharT, _Traits>&
+  operator>>(::std::basic_istream<_CharT, _Traits>& __is, cauchy_distribution& __x)
+  {
+    using istream_type                        = ::std::basic_istream<_CharT, _Traits>;
+    using ios_base                            = typename istream_type::ios_base;
+    using result_type                         = typename cauchy_distribution::result_type;
+    using param_type                          = typename cauchy_distribution::param_type;
+    const typename ios_base::fmtflags __flags = __is.flags();
+    __is.flags(ios_base::dec | ios_base::skipws);
+    result_type __a;
+    result_type __b;
+    __is >> __a >> __b;
+    if (!__is.fail())
+    {
+      __x.param(param_type{__a, __b});
+    }
+    __is.flags(__flags);
+    return __is;
+  }
+#endif // !_CCCL_COMPILER(NVRTC)
+};
+
+_CCCL_END_NAMESPACE_CUDA_STD
+
+#include <cuda/std/__cccl/epilogue.h>
+
+#endif // _CUDA_STD___RANDOM_CAUCHY_DISTRIBUTION_H

libcudacxx/include/cuda/std/__random_

@@ -22,6 +22,7 @@
 #endif // no system header
 
 #include <cuda/std/__random/bernoulli_distribution.h>
+#include <cuda/std/__random/cauchy_distribution.h>
 #include <cuda/std/__random/exponential_distribution.h>
 #include <cuda/std/__random/linear_congruential_engine.h>
 #include <cuda/std/__random/normal_distribution.h>

libcudacxx/test/libcudacxx/std/random/distribution/cauchy.pass.cpp

@@ -0,0 +1,53 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+// SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+//
+// REQUIRES: long_tests
+
+// <random>
+
+// template<class RealType = double>
+// class cauchy_distribution
+
+#include <cuda/std/__random_>
+#include <cuda/std/cassert>
+#include <cuda/std/cmath>
+#include <cuda/std/numbers>
+
+#include "random_utilities/test_distribution.h"
+#include "test_macros.h"
+
+template <class T>
+struct cauchy_cdf
+{
+  using P = typename cuda::std::cauchy_distribution<T>::param_type;
+
+  __host__ __device__ double operator()(double x, const P& p) const
+  {
+    // CDF of Cauchy distribution: F(x; a, b) = (1/π) * arctan((x - a) / b) + 0.5
+    return (1.0 / cuda::std::numbers::pi) * cuda::std::atan((x - p.a()) / p.b()) + 0.5;
+  }
+};
+
+template <class T>
+__host__ __device__ void test()
+{
+  [[maybe_unused]] const bool test_constexpr = false;
+  using D                                    = cuda::std::cauchy_distribution<T>;
+  using P                                    = typename D::param_type;
+  using G                                    = cuda::std::philox4x64;
+  cuda::std::array<P, 5> params              = {P(0, 1), P(10, 2), P(-5, 0.5), P(4, 5), P(1000, 100)};
+  test_distribution<D, true, G, test_constexpr>(params, cauchy_cdf<T>{});
+}
+
+int main(int, char**)
+{
+  test<double>();
+  test<float>();
+  return 0;
+}