// This file is part of Eigen, a lightweight C++ template library
// for linear algebra.
//
// Copyright (C) 2008 Gael Guennebaud <gael.guennebaud@inria.fr>
// Copyright (C) 2007-2011 Benoit Jacob <jacob.benoit.1@gmail.com>
//
// This Source Code Form is subject to the terms of the Mozilla
// Public License v. 2.0. If a copy of the MPL was not distributed
// with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
// SPDX-License-Identifier: MPL-2.0

#ifndef EIGEN_CORE_MODULE_H
#define EIGEN_CORE_MODULE_H

// Eigen version information.
#include "Version"

// first thing Eigen does: stop the compiler from reporting useless warnings.
#include "src/Core/util/DisableStupidWarnings.h"

// then include this file where all our macros are defined. It's really important to do it first because
// it's where we do all the compiler/OS/arch detections and define most defaults.
#include "src/Core/util/Macros.h"

// This detects SSE/AVX/NEON/etc. and configure alignment settings
#include "src/Core/util/ConfigureVectorization.h"

// We need cuda_runtime.h/hip_runtime.h to ensure that
// the EIGEN_USING_STD macro works properly on the device side
#if defined(EIGEN_CUDACC)
#include <cuda_runtime.h>
#elif defined(EIGEN_HIPCC)
#include <hip/hip_runtime.h>
#endif

#if defined(EIGEN_EXCEPTIONS) || defined(EIGEN_GPU_COMPILE_PHASE)
#include <new>
#endif

// Prevent ICC from specializing std::complex operators that silently fail
// on device. This allows us to use our own device-compatible specializations
// instead.
#if EIGEN_COMP_ICC && defined(EIGEN_GPU_COMPILE_PHASE) && !defined(_OVERRIDE_COMPLEX_SPECIALIZATION_)
#define _OVERRIDE_COMPLEX_SPECIALIZATION_ 1
#endif
#include <complex>

// this include file manages BLAS and MKL related macros
// and inclusion of their respective header files
#include "src/Core/util/MKL_support.h"
#include "src/Core/util/AOCL_Support.h"

// EIGEN_HAS_GPU_FP16 is defined in Macros.h when compiling with CUDA or HIP (excluding SPIR-V).
// TODO: Remove EIGEN_HAS_GPU_BF16 similarly once HIP bf16 guards are cleaned up.

#if defined(EIGEN_HAS_CUDA_BF16) || defined(EIGEN_HAS_HIP_BF16)
#define EIGEN_HAS_GPU_BF16
#endif

#if (defined _OPENMP) && (!defined EIGEN_DONT_PARALLELIZE)
#define EIGEN_HAS_OPENMP
#endif

#ifdef EIGEN_HAS_OPENMP
#include <atomic>
#include <omp.h>
#endif

#if !EIGEN_COMP_ARM
#define EIGEN_HAS_ERRNO
#endif

#ifdef EIGEN_HAS_ERRNO
#include <cerrno>
#endif
#include <cstddef>
#include <cstdlib>
#include <cmath>
#include <functional>
#ifndef EIGEN_NO_IO
#include <sstream>
#include <iosfwd>
#endif
#include <cstring>
#include <string>
#include <limits>
#include <climits>  // for CHAR_BIT
// for min/max:
#include <algorithm>

#include <array>
#include <memory>
#include <vector>

// for std::is_nothrow_move_assignable
#include <type_traits>

// for std::move, std::forward, std::declval
#include <utility>

// for std::this_thread::yield().
#if !defined(EIGEN_USE_BLAS) && (defined(EIGEN_HAS_OPENMP) || defined(EIGEN_GEMM_THREADPOOL))
#include <thread>
#endif

// for __cpp_lib feature test macros
#if defined(__has_include) && __has_include(<version>)
#include <version>
#endif

// for std::bit_cast()
#if defined(__cpp_lib_bit_cast) && __cpp_lib_bit_cast >= 201806L
#include <bit>
#endif

// for outputting debug info
#ifdef EIGEN_DEBUG_ASSIGN
#include <iostream>
#endif

// required for __cpuid, needs to be included after cmath
// also required for _BitScanReverse on Windows on ARM
#if EIGEN_COMP_MSVC && (EIGEN_ARCH_i386_OR_x86_64 || EIGEN_ARCH_ARM64)
#include <intrin.h>
#endif

// Required for querying cache sizes on Linux and macOS.
#if EIGEN_OS_LINUX
#if !defined(EIGEN_NO_CPU_CACHE_SYSFS)
#include <cstdio>
#include <fcntl.h>
#include <sched.h>
#endif
#include <unistd.h>
#elif EIGEN_OS_MAC
#include <sys/types.h>
#include <sys/sysctl.h>
#endif

#if defined(EIGEN_USE_SYCL)
#undef min
#undef max
#undef isnan
#undef isinf
#undef isfinite
#include <CL/sycl.hpp>
#include <map>
#include <thread>
#ifndef EIGEN_SYCL_LOCAL_THREAD_DIM0
#define EIGEN_SYCL_LOCAL_THREAD_DIM0 16
#endif
#ifndef EIGEN_SYCL_LOCAL_THREAD_DIM1
#define EIGEN_SYCL_LOCAL_THREAD_DIM1 16
#endif
#endif

namespace Eigen {

using std::ptrdiff_t;
using std::size_t;

}  // namespace Eigen

/** \defgroup Core_Module Core module
 * This is the main module of Eigen providing dense matrix and vector support
 * (both fixed and dynamic size) with all the features corresponding to a BLAS library
 * and much more...
 *
 * \code
 * #include <Eigen/Core>
 * \endcode
 */

#ifdef EIGEN_USE_LAPACKE
#ifdef EIGEN_USE_MKL
#include "mkl_lapacke.h"
#elif defined(EIGEN_LAPACKE_SYSTEM)
#include <lapacke.h>
#else
#include "src/LAPACKESupport/lapacke.h"
#endif
#endif

// IWYU pragma: begin_exports
// Device overloads of the std::complex operators. They must precede every template that applies an operator to a
// dependent Scalar; see the header.
#ifdef EIGEN_GPUCC
#include "src/Core/arch/GPU/Complex.h"
#endif
#include "src/Core/util/Constants.h"
#include "src/Core/util/Meta.h"
#include "src/Core/util/Assert.h"
#include "src/Core/util/ForwardDeclarations.h"
#include "src/Core/util/StaticAssert.h"
#include "src/Core/util/XprHelper.h"
#include "src/Core/util/ShapeChecks.h"
#include "src/Core/util/Memory.h"
#include "src/Core/util/IntegralConstant.h"
#include "src/Core/util/Serializer.h"
#include "src/Core/util/SymbolicIndex.h"
#include "src/Core/util/EmulateArray.h"
#include "src/Core/util/MoreMeta.h"
#include "src/Core/util/LruCache.h"

#include "src/Core/NumTraits.h"
#include "src/Core/MathFunctions.h"
#include "src/Core/RandomImpl.h"
#include "src/Core/GenericPacketMath.h"
#include "src/Core/MathFunctionsImpl.h"
#include "src/Core/arch/Default/ConjHelper.h"
// Generic half float support
#include "src/Core/arch/Default/Half.h"
#include "src/Core/arch/Default/BFloat16.h"
#include "src/Core/arch/Default/GenericPacketMathFunctionsFwd.h"

#if defined(EIGEN_VECTORIZE_GENERIC) && !defined(EIGEN_DONT_VECTORIZE)
#include "src/Core/arch/clang/PacketMath.h"
#include "src/Core/arch/clang/TypeCasting.h"
#include "src/Core/arch/clang/Complex.h"
#include "src/Core/arch/clang/Reductions.h"
#include "src/Core/arch/clang/MathFunctions.h"
#else
#if defined EIGEN_VECTORIZE_AVX512
#include "src/Core/arch/SSE/PacketMath.h"
#include "src/Core/arch/SSE/Reductions.h"
#include "src/Core/arch/AVX/PacketMath.h"
#include "src/Core/arch/AVX/Reductions.h"
#include "src/Core/arch/AVX512/PacketMath.h"
#include "src/Core/arch/AVX512/Reductions.h"
#if defined EIGEN_VECTORIZE_AVX512FP16
#include "src/Core/arch/AVX512/PacketMathFP16.h"
#endif
#include "src/Core/arch/SSE/TypeCasting.h"
#include "src/Core/arch/AVX/TypeCasting.h"
#include "src/Core/arch/AVX512/TypeCasting.h"
#if defined EIGEN_VECTORIZE_AVX512FP16
#include "src/Core/arch/AVX512/TypeCastingFP16.h"
#endif
#include "src/Core/arch/SSE/Complex.h"
#include "src/Core/arch/AVX/Complex.h"
#include "src/Core/arch/AVX512/Complex.h"
#include "src/Core/arch/SSE/MathFunctions.h"
#include "src/Core/arch/AVX/MathFunctions.h"
#include "src/Core/arch/AVX512/MathFunctions.h"
#if defined EIGEN_VECTORIZE_AVX512FP16
#include "src/Core/arch/AVX512/MathFunctionsFP16.h"
#endif
#include "src/Core/arch/AVX512/TrsmKernel.h"
#elif defined EIGEN_VECTORIZE_AVX
// Use AVX for floats and doubles, SSE for integers
#include "src/Core/arch/SSE/PacketMath.h"
#include "src/Core/arch/SSE/Reductions.h"
#include "src/Core/arch/SSE/TypeCasting.h"
#include "src/Core/arch/SSE/Complex.h"
#include "src/Core/arch/AVX/PacketMath.h"
#include "src/Core/arch/AVX/Reductions.h"
#include "src/Core/arch/AVX/TypeCasting.h"
#include "src/Core/arch/AVX/Complex.h"
#include "src/Core/arch/SSE/MathFunctions.h"
#include "src/Core/arch/AVX/MathFunctions.h"
#elif defined EIGEN_VECTORIZE_SSE
#include "src/Core/arch/SSE/PacketMath.h"
#include "src/Core/arch/SSE/Reductions.h"
#include "src/Core/arch/SSE/TypeCasting.h"
#include "src/Core/arch/SSE/MathFunctions.h"
#include "src/Core/arch/SSE/Complex.h"
#endif

#if defined(EIGEN_VECTORIZE_ALTIVEC) || defined(EIGEN_VECTORIZE_VSX)
#include "src/Core/arch/AltiVec/PacketMath.h"
#include "src/Core/arch/AltiVec/TypeCasting.h"
#include "src/Core/arch/AltiVec/MathFunctions.h"
#include "src/Core/arch/AltiVec/Complex.h"
#elif defined EIGEN_VECTORIZE_NEON
#include "src/Core/arch/NEON/PacketMath.h"
#include "src/Core/arch/NEON/TypeCasting.h"
#include "src/Core/arch/NEON/MathFunctions.h"
#include "src/Core/arch/NEON/Complex.h"
#elif defined EIGEN_VECTORIZE_SME
// Keep NEON packet types outside the specialized streaming kernels.
#include "src/Core/arch/NEON/PacketMath.h"
#include "src/Core/arch/NEON/TypeCasting.h"
#include "src/Core/arch/NEON/MathFunctions.h"
#include "src/Core/arch/NEON/Complex.h"
#include "src/Core/arch/SME/PacketMath.h"
#elif defined EIGEN_VECTORIZE_LSX
#include "src/Core/arch/LSX/PacketMath.h"
#include "src/Core/arch/LSX/TypeCasting.h"
#include "src/Core/arch/LSX/MathFunctions.h"
#include "src/Core/arch/LSX/Complex.h"
#elif defined EIGEN_VECTORIZE_SVE
#include "src/Core/arch/SVE/PacketMath.h"
#include "src/Core/arch/SVE/TypeCasting.h"
#include "src/Core/arch/SVE/MathFunctions.h"
#include "src/Core/arch/SVE/Complex.h"
#elif defined EIGEN_VECTORIZE_RVV10
#include "src/Core/arch/RVV10/PacketMathDecl.h"
#include "src/Core/arch/RVV10/PacketMath.h"
#include "src/Core/arch/RVV10/PacketMath2.h"
#include "src/Core/arch/RVV10/PacketMath4.h"
#include "src/Core/arch/RVV10/TypeCasting.h"
#include "src/Core/arch/RVV10/MathFunctions.h"
#if defined EIGEN_VECTORIZE_RVV10FP16
#include "src/Core/arch/RVV10/PacketMathFP16.h"
#endif
#if defined EIGEN_VECTORIZE_RVV10BF16
#include "src/Core/arch/RVV10/PacketMathBF16.h"
#endif
#if EIGEN_RISCV64_DEFAULT_LMUL == 1
#include "src/Core/arch/RVV10/Complex.h"
#include "src/Core/arch/RVV10/Complex2.h"
#endif
#elif defined EIGEN_VECTORIZE_ZVECTOR
#include "src/Core/arch/ZVector/PacketMath.h"
#include "src/Core/arch/ZVector/MathFunctions.h"
#include "src/Core/arch/ZVector/Complex.h"
#elif defined EIGEN_VECTORIZE_MSA
#include "src/Core/arch/MSA/PacketMath.h"
#include "src/Core/arch/MSA/MathFunctions.h"
#include "src/Core/arch/MSA/Complex.h"
#elif defined EIGEN_VECTORIZE_HVX
#include "src/Core/arch/HVX/PacketMath.h"
#endif

#if defined EIGEN_VECTORIZE_GPU
#include "src/Core/arch/GPU/PacketMath.h"
#include "src/Core/arch/GPU/MathFunctions.h"
#include "src/Core/arch/GPU/TypeCasting.h"
#endif

#if defined(EIGEN_USE_SYCL)
#include "src/Core/arch/SYCL/InteropHeaders.h"
#if !defined(EIGEN_DONT_VECTORIZE_SYCL)
#include "src/Core/arch/SYCL/PacketMath.h"
#include "src/Core/arch/SYCL/MathFunctions.h"
#include "src/Core/arch/SYCL/TypeCasting.h"
#endif
#endif

#endif  // #ifndef EIGEN_VECTORIZE_GENERIC

#include "src/Core/arch/Default/Settings.h"
// This file provides generic implementations valid for scalar as well
#include "src/Core/arch/Default/GenericPacketMathFunctions.h"

#include "src/Core/functors/TernaryFunctors.h"
#include "src/Core/functors/BinaryFunctors.h"
#include "src/Core/functors/UnaryFunctors.h"
#include "src/Core/functors/NullaryFunctors.h"
#include "src/Core/functors/StlFunctors.h"
#include "src/Core/functors/AssignmentFunctors.h"

// Specializations of vectorized activation functions for NEON.
#ifdef EIGEN_VECTORIZE_NEON
#include "src/Core/arch/NEON/UnaryFunctors.h"
#endif

#include "src/Core/util/IndexedViewHelper.h"
#include "src/Core/util/ReshapedHelper.h"
#include "src/Core/ArithmeticSequence.h"
#ifndef EIGEN_NO_IO
#include "src/Core/IO.h"
#endif
#include "src/Core/DenseCoeffsBase.h"
#include "src/Core/DenseBase.h"
#include "src/Core/MatrixBase.h"
#include "src/Core/EigenBase.h"

#include "src/Core/Product.h"
#include "src/Core/CoreEvaluators.h"
#include "src/Core/AssignEvaluator.h"
#include "src/Core/RealView.h"
#include "src/Core/Assign.h"

#include "src/Core/ArrayBase.h"
#include "src/Core/util/BlasUtil.h"
#include "src/Core/DenseStorage.h"
#include "src/Core/NestByValue.h"

#include "src/Core/ReturnByValue.h"
#include "src/Core/NoAlias.h"
#include "src/Core/PlainObjectBase.h"
#include "src/Core/Matrix.h"
#include "src/Core/Array.h"
#include "src/Core/StructuredBindings.h"
#include "src/Core/Fill.h"
#include "src/Core/CwiseTernaryOp.h"
#include "src/Core/CwiseBinaryOp.h"
#include "src/Core/CwiseUnaryOp.h"
#include "src/Core/CwiseNullaryOp.h"
#include "src/Core/CwiseUnaryView.h"
#include "src/Core/SelfCwiseBinaryOp.h"
#include "src/Core/InnerProduct.h"
#include "src/Core/SafeScaling.h"
#include "src/Core/Dot.h"
#include "src/Core/StableNorm.h"
#include "src/Core/Stride.h"
#include "src/Core/MapBase.h"
#include "src/Core/Map.h"
#include "src/Core/Ref.h"
#include "src/Core/Block.h"
#include "src/Core/VectorBlock.h"
#include "src/Core/IndexedView.h"
#include "src/Core/Reshaped.h"
#include "src/Core/Transpose.h"
#include "src/Core/DiagonalMatrix.h"
#include "src/Core/Diagonal.h"
#include "src/Core/DiagonalProduct.h"
#include "src/Core/SkewSymmetricMatrix3.h"
#include "src/Core/Redux.h"
#include "src/Core/Visitor.h"
#include "src/Core/Predicator.h"
#include "src/Core/FindCoeff.h"
#include "src/Core/ApproxComparisons.h"
#include "src/Core/Swap.h"
#include "src/Core/CommaInitializer.h"
#include "src/Core/GeneralProduct.h"
#include "src/Core/Solve.h"
#include "src/Core/Inverse.h"
#include "src/Core/SolverBase.h"
#include "src/Core/PermutationMatrix.h"
#include "src/Core/Transpositions.h"
#include "src/Core/TriangularMatrix.h"
#include "src/Core/MatrixNorms.h"
#include "src/Core/SelfAdjointView.h"
#include "src/Core/products/GeneralBlockPanelKernel.h"
#include "src/Core/DeviceWrapper.h"
#ifdef EIGEN_GEMM_THREADPOOL
#include "ThreadPool"
#endif
#include "src/Core/products/Parallelizer.h"
#include "src/Core/ProductEvaluators.h"
#include "src/Core/ScaledPermutationMatrix.h"
#include "src/Core/products/GeneralMatrixVector.h"
#include "src/Core/products/GeneralMatrixMatrix.h"
#include "src/Core/SolveTriangular.h"
#include "src/Core/products/GeneralMatrixMatrixTriangular.h"
#include "src/Core/products/SelfadjointMatrixVector.h"
#include "src/Core/products/SelfadjointMatrixMatrix.h"
#include "src/Core/products/SelfadjointProduct.h"
#include "src/Core/products/SelfadjointRank2Update.h"
#include "src/Core/products/TriangularMatrixVector.h"
#include "src/Core/products/TriangularMatrixMatrix.h"
#include "src/Core/products/TriangularSolverMatrix.h"
#include "src/Core/products/TriangularSolverVector.h"
#include "src/Core/TriangularInPlace.h"
#include "src/Core/BandMatrix.h"
#include "src/Core/CoreIterators.h"
#include "src/Core/ConditionEstimator.h"

#if !defined(EIGEN_VECTORIZE_GENERIC)
#if defined(EIGEN_VECTORIZE_VSX)
#include "src/Core/arch/AltiVec/MatrixProduct.h"
#elif defined EIGEN_VECTORIZE_NEON
#include "src/Core/arch/NEON/GeneralBlockPanelKernel.h"
#elif defined EIGEN_VECTORIZE_SVE
#include "src/Core/arch/SVE/GeneralBlockPanelKernel.h"
#elif defined EIGEN_VECTORIZE_SME
#include "src/Core/arch/SME/GeneralBlockPanelKernel.h"
#include "src/Core/arch/SME/VectorKernels.h"
#elif defined EIGEN_VECTORIZE_LSX
#include "src/Core/arch/LSX/GeneralBlockPanelKernel.h"
#elif defined EIGEN_VECTORIZE_RVV10
#include "src/Core/arch/RVV10/GeneralBlockPanelKernel.h"
#endif

#if defined(EIGEN_VECTORIZE_AVX512)
#include "src/Core/arch/AVX512/GemmKernel.h"
#endif
#endif

#include "src/Core/Select.h"
#include "src/Core/VectorwiseOp.h"
#include "src/Core/PartialReduxEvaluator.h"
#include "src/Core/Random.h"
#include "src/Core/Replicate.h"
#include "src/Core/ConcatOp.h"
#include "src/Core/Reverse.h"
#include "src/Core/ArrayWrapper.h"
#include "src/Core/StlIterators.h"

#ifdef EIGEN_USE_BLAS
#include "src/Core/products/GeneralMatrixMatrix_BLAS.h"
#include "src/Core/products/GeneralMatrixVector_BLAS.h"
#include "src/Core/products/GeneralMatrixMatrixTriangular_BLAS.h"
#include "src/Core/products/SelfadjointMatrixMatrix_BLAS.h"
#include "src/Core/products/SelfadjointMatrixVector_BLAS.h"
#include "src/Core/products/TriangularMatrixMatrix_BLAS.h"
#include "src/Core/products/TriangularMatrixVector_BLAS.h"
#include "src/Core/products/TriangularSolverMatrix_BLAS.h"
#endif  // EIGEN_USE_BLAS

#ifdef EIGEN_USE_MKL_VML
#include "src/Core/Assign_MKL.h"
#endif

#ifdef EIGEN_USE_AOCL_VML
#include "src/Core/Assign_AOCL.h"
#endif

#include "src/Core/GlobalFunctions.h"
// IWYU pragma: end_exports

#include "src/Core/util/ReenableStupidWarnings.h"

#endif  // EIGEN_CORE_MODULE_H