BandMatCluster.hpp

Go to the documentation of this file.

/*******************************************************************************
 * Copyright (C) 2017-2026 Theodore Chang
 *
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 ******************************************************************************/
// ReSharper disable CppCStyleCast
#ifndef BANDMATCLUSTER_HPP
#define BANDMATCLUSTER_HPP
 
#include "../DenseMat.hpp"
 
#include <ezp/ezp/pgbsv.hpp>
 
template<sp_d T> class BandMatCluster : public DenseMat<T> {
    using solver_t = ezp::pgbsv<T, la_it>;
    using indexer_t = solver_t::indexer;
 
    static T bin;
 
    solver_t solver;
    indexer_t indexer;
 
    int solve_trs(Mat<T>&, Mat<T>&&);
 
protected:
    const uword l_band, u_band;
 
    using DenseMat<T>::direct_solve;
 
    int direct_solve(Mat<T>&, Mat<T>&&) override;
 
public:
    BandMatCluster(const uword in_size, const uword in_l, const uword in_u)
        : DenseMat<T>(in_size, in_size, (2 * (in_l + in_u) + 1) * in_size)
        , solver()
        , indexer(in_size, in_l, in_u)
        , l_band(in_l)
        , u_band(in_u) {
        if(2 * (in_l + in_u) + 1 >= in_size)
            suanpan_warning("The storage requirement for the banded matrix is larger than that of a full matrix, consider using a full/sparse matrix instead.\n");
    }
 
    unique_ptr<MetaMat<T>> unique_copy() override { return std::make_unique<BandMatCluster>(*this); }
 
    void nullify(const uword K) override {
        this->factored = false;
        suanpan::for_each(std::max(K, u_band) - u_band, std::min(this->n_rows, K + l_band + 1), [&](const uword I) { this->memory[2 * u_band + l_band + I + 2 * K * (l_band + u_band)] = T(0); });
        suanpan::for_each(std::max(K, l_band) - l_band, std::min(this->n_cols, K + u_band + 1), [&](const uword I) { this->memory[2 * u_band + l_band + K + 2 * I * (l_band + u_band)] = T(0); });
    }
 
    T operator()(const uword in_row, const uword in_col) const override {
        const auto pos = indexer(in_row, in_col);
        if(pos < 0) [[unlikely]]
            return bin = T(0);
        return this->memory[pos];
    }
 
    T& unsafe_at(const uword in_row, const uword in_col) override {
        this->factored = false;
        return this->memory[indexer(in_row, in_col)];
    }
 
    T& at(const uword in_row, const uword in_col) override {
        const auto pos = indexer(in_row, in_col);
        if(pos < 0) [[unlikely]]
            return bin = T(0);
        this->factored = false;
        return this->memory[pos];
    }
 
    Mat<T> operator*(const Mat<T>&) const override;
};
 
template<sp_d T> T BandMatCluster<T>::bin = T(0);
 
template<sp_d T> Mat<T> BandMatCluster<T>::operator*(const Mat<T>& X) const {
    static constexpr auto TRAN = 'N';
    static constexpr blas_int INC = 1;
    static constexpr T ALPHA{1}, BETA{0};
 
    Mat<T> Y(arma::size(X));
 
    const auto s_band = l_band + u_band;
 
    const auto M = static_cast<blas_int>(this->n_rows);
    const auto N = static_cast<blas_int>(this->n_cols);
    const auto KL = static_cast<blas_int>(l_band);
    const auto KU = static_cast<blas_int>(u_band);
    const auto LDA = static_cast<blas_int>(2 * s_band + 1);
 
    if constexpr(std::is_same_v<T, float>) suanpan::for_each(X.n_cols, [&](const uword I) { arma_fortran(arma_sgbmv)(&TRAN, &M, &N, &KL, &KU, &ALPHA, this->memptr() + s_band, &LDA, X.colptr(I), &INC, &BETA, Y.colptr(I), &INC); });
    else suanpan::for_each(X.n_cols, [&](const uword I) { arma_fortran(arma_dgbmv)(&TRAN, &M, &N, &KL, &KU, &ALPHA, this->memptr() + s_band, &LDA, X.colptr(I), &INC, &BETA, Y.colptr(I), &INC); });
 
    return Y;
}
 
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wnarrowing"
template<sp_d T> int BandMatCluster<T>::direct_solve(Mat<T>& X, Mat<T>&& B) {
    if(this->factored) return this->solve_trs(X, std::move(B));
 
    suanpan_assert([&] { if(this->n_rows != this->n_cols) throw std::invalid_argument("requires a square matrix"); });
 
    this->factored = true;
 
    const auto INFO = bcast_from_root(solver.solve({this->n_rows, this->n_cols, this->l_band, this->u_band, this->memptr()}, {B.n_rows, B.n_cols, B.memptr()}));
 
    if(0 == INFO) bcast_from_root(X = std::move(B));
    else suanpan_error("Error code {} received, the matrix is probably singular.\n", INFO);
 
    return INFO;
}
 
template<sp_d T> int BandMatCluster<T>::solve_trs(Mat<T>& X, Mat<T>&& B) {
    const auto INFO = bcast_from_root(solver.solve({B.n_rows, B.n_cols, B.memptr()}));
 
    if(0 == INFO) bcast_from_root(X = std::move(B));
    else suanpan_error("Error code {} received, the matrix is probably singular.\n", INFO);
 
    return INFO;
}
#pragma GCC diagnostic pop
 
#endif