//@HEADER // ************************************************************************ // // Kokkos v. 4.0 // Copyright (2022) National Technology & Engineering // Solutions of Sandia, LLC (NTESS). // // Under the terms of Contract DE-NA0003525 with NTESS, // the U.S. Government retains certain rights in this software. // // Part of Kokkos, under the Apache License v2.0 with LLVM Exceptions. // See https://kokkos.org/LICENSE for license information. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // //@HEADER #ifndef KOKKOSBATCHED_RIGHT_EIGENVECTOR_FROM_SCHUR_SERIAL_INTERNAL_HPP #define KOKKOSBATCHED_RIGHT_EIGENVECTOR_FROM_SCHUR_SERIAL_INTERNAL_HPP /// \author Kyungjoo Kim (kyukim@sandia.gov) #include "KokkosBatched_Util.hpp" #include "KokkosBatched_ShiftedTrsv_Serial_Internal.hpp" namespace KokkosBatched { /// /// Serial Internal Impl /// ==================== /// /// this impl follows the flame interface of householder transformation /// struct SerialRightEigenvectorFromSchurInternal { /// Given a quasi upper triangular matrix S (m x m), this computes all right /// eigenvectors. /// /// Parameters: /// [in]m /// A dimension of the square matrix S. /// [in]S, [in]ss0, [in]ss1 /// A quasi upper triangular part of Schur decomposition which is computed /// A = U^H S U /// [out]V, [in]vs0, [out]vs1 /// A set of right eigen vectors. /// [in]w /// contiguous workspace that can hold complex array (m) template KOKKOS_INLINE_FUNCTION static int invoke(const int m, /* */ ValueType *S, const int ss0, const int ss1, /* */ ValueType *V, const int vs0, const int vs1, /* */ ValueType *w, const int *blks) { typedef ValueType value_type; typedef Kokkos::ArithTraits ats; // typedef typename ats::mag_type mag_type; typedef Kokkos::complex complex_type; const value_type zero(0), one(1); // const int ss(ss0+ss1); /// SerialSetInternal::invoke(m, m, zero, V, vs0, vs1); value_type *b = w; // consider complex case /// partitions used for loop iteration Partition2x2 S_part2x2(ss0, ss1); Partition3x3 S_part3x3(ss0, ss1); Partition1x2 V_part1x2(vs1); Partition1x3 V_part1x3(vs1); /// initial partition of S where ABR has a zero dimension S_part2x2.partWithABR(S, m, m, 0, 0); V_part1x2.partWithAR(V, m, 0); // const mag_type tol = ats::epsilon(); int m_stl = m; for (; m_stl > 0;) { /// part 2x2 into 3x3 const int mA11 = blks[m_stl - 1]; assert(((mA11 == 1) || (mA11 == 2)) && "RightEigenvectorFromSchur: blk is not 1x1 nor 2x2"); S_part3x3.partWithATL(S_part2x2, mA11, mA11); V_part1x3.partWithAL(V_part1x2, mA11); const int m_stl_minus_mA11 = m_stl - mA11; if (mA11 == 1) { /// real eigenvalue const value_type lambda = *S_part3x3.A11; /// initialize a right eigen vector for (int i = 0; i < m_stl_minus_mA11; ++i) b[i] = -S_part3x3.A01[i * ss0]; b[m_stl - 1] = one; /// perform shifted trsv SerialShiftedTrsvInternalUpper::invoke(m_stl_minus_mA11, lambda, S_part3x3.A00, ss0, ss1, w, 1, blks); /// copy back to V for (int i = 0; i < m_stl; ++i) V_part1x3.A1[i * vs0] = w[i]; for (int i = m_stl; i < m; ++i) V_part1x3.A1[i * vs0] = zero; } else { /// complex eigen pair const value_type alpha11 = S_part3x3.A11[0], alpha12 = S_part3x3.A11[ss1], alpha21 = S_part3x3.A11[ss0], beta = ats::sqrt(-alpha12 * alpha21); const complex_type lambda(alpha11, beta); complex_type *bc = (complex_type *)(b); /// initialize a right eigen vector const value_type *S_A01_a = S_part3x3.A01; const value_type *S_A01_b = S_part3x3.A01 + ss1; for (int i = 0; i < m_stl_minus_mA11; ++i) bc[i] = complex_type(-S_A01_a[i * ss0] * beta, S_A01_b[i * ss0] * alpha21); bc[m_stl - 2] = complex_type(beta, zero); bc[m_stl - 1] = complex_type(zero, -alpha21); /// perform shifted trsv SerialShiftedTrsvInternalUpper::invoke(m_stl_minus_mA11, lambda, S_part3x3.A00, ss0, ss1, bc, 1, blks); /// copy back to V value_type *V_A1_r = V_part1x3.A1; value_type *V_A1_i = V_part1x3.A1 + vs1; for (int i = 0; i < m_stl; ++i) { V_A1_r[i * vs0] = bc[i].real(); V_A1_i[i * vs0] = bc[i].imag(); } for (int i = m_stl; i < m; ++i) { V_A1_r[i * vs0] = zero; V_A1_i[i * vs0] = zero; } /// --------------------------------------------------- } S_part2x2.mergeToABR(S_part3x3); V_part1x2.mergeToAR(V_part1x3); m_stl -= mA11; } return 0; } }; } // namespace KokkosBatched #endif