Files
armadillo-code/include/armadillo_bits/op_vectorise_meat.hpp
T

453 lines
8.6 KiB
C++

// SPDX-License-Identifier: Apache-2.0
//
// Copyright 2008-2016 Conrad Sanderson (http://conradsanderson.id.au)
// Copyright 2008-2016 National ICT Australia (NICTA)
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
// ------------------------------------------------------------------------
//! \addtogroup op_vectorise
//! @{
template<typename T1>
inline
void
op_vectorise_col::apply(Mat<typename T1::elem_type>& out, const Op<T1,op_vectorise_col>& in)
{
arma_debug_sigprint();
op_vectorise_col::apply_direct(out, in.m);
}
template<typename T1>
inline
void
op_vectorise_col::apply_direct(Mat<typename T1::elem_type>& out, const T1& expr)
{
arma_debug_sigprint();
typedef typename T1::elem_type eT;
// allow detection of in-place operation
if(is_Mat<T1>::value)
{
const unwrap<T1> U(expr);
if(&out == &(U.M))
{
// output matrix is the same as the input matrix
out.set_size(out.n_elem, 1); // set_size() doesn't destroy data as long as the number of elements in the matrix remains the same
}
else
{
out.set_size(U.M.n_elem, 1);
arrayops::copy(out.memptr(), U.M.memptr(), U.M.n_elem);
}
}
else
if(is_subview<T1>::value)
{
const subview<eT>& sv = reinterpret_cast< const subview<eT>& >(expr);
if(&out == &(sv.m))
{
Mat<eT> tmp;
op_vectorise_col::apply_subview(tmp, sv);
out.steal_mem(tmp);
}
else
{
op_vectorise_col::apply_subview(out, sv);
}
}
else
if((is_Mat<typename Proxy<T1>::stored_type>::value) || (arma_config::openmp && Proxy<T1>::use_mp))
{
const quasi_unwrap<T1> U(expr);
if(U.is_alias(out))
{
Mat<eT> tmp(U.M.memptr(), U.M.n_elem, 1);
out.steal_mem(tmp);
}
else
{
out.set_size(U.M.n_elem, 1);
arrayops::copy(out.memptr(), U.M.memptr(), U.M.n_elem);
}
}
else
{
const Proxy<T1> P(expr);
if(P.is_alias(out))
{
Mat<eT> tmp;
op_vectorise_col::apply_proxy(tmp, P);
out.steal_mem(tmp);
}
else
{
op_vectorise_col::apply_proxy(out, P);
}
}
}
template<typename eT>
inline
void
op_vectorise_col::apply_subview(Mat<eT>& out, const subview<eT>& sv)
{
arma_debug_sigprint();
const uword sv_n_rows = sv.n_rows;
const uword sv_n_cols = sv.n_cols;
out.set_size(sv.n_elem, 1);
eT* out_mem = out.memptr();
for(uword col=0; col < sv_n_cols; ++col)
{
arrayops::copy(out_mem, sv.colptr(col), sv_n_rows);
out_mem += sv_n_rows;
}
}
template<typename T1>
inline
void
op_vectorise_col::apply_proxy(Mat<typename T1::elem_type>& out, const Proxy<T1>& P)
{
arma_debug_sigprint();
typedef typename T1::elem_type eT;
const uword N = P.get_n_elem();
out.set_size(N, 1);
eT* outmem = out.memptr();
if(Proxy<T1>::use_at == false)
{
// TODO: add handling of aligned access ?
typename Proxy<T1>::ea_type A = P.get_ea();
uword i,j;
for(i=0, j=1; j < N; i+=2, j+=2)
{
const eT tmp_i = A[i];
const eT tmp_j = A[j];
outmem[i] = tmp_i;
outmem[j] = tmp_j;
}
if(i < N)
{
outmem[i] = A[i];
}
}
else
{
const uword n_rows = P.get_n_rows();
const uword n_cols = P.get_n_cols();
if(n_rows == 1)
{
for(uword i=0; i < n_cols; ++i)
{
outmem[i] = P.at(0,i);
}
}
else
{
for(uword col=0; col < n_cols; ++col)
for(uword row=0; row < n_rows; ++row)
{
*outmem = P.at(row,col);
outmem++;
}
}
}
}
template<typename T1>
inline
void
op_vectorise_row::apply(Mat<typename T1::elem_type>& out, const Op<T1,op_vectorise_row>& in)
{
arma_debug_sigprint();
op_vectorise_row::apply_direct(out, in.m);
}
template<typename T1>
inline
void
op_vectorise_row::apply_direct(Mat<typename T1::elem_type>& out, const T1& expr)
{
arma_debug_sigprint();
typedef typename T1::elem_type eT;
const Proxy<T1> P(expr);
if(P.is_alias(out))
{
Mat<eT> tmp;
op_vectorise_row::apply_proxy(tmp, P);
out.steal_mem(tmp);
}
else
{
op_vectorise_row::apply_proxy(out, P);
}
}
template<typename T1>
inline
void
op_vectorise_row::apply_proxy(Mat<typename T1::elem_type>& out, const Proxy<T1>& P)
{
arma_debug_sigprint();
typedef typename T1::elem_type eT;
const uword n_rows = P.get_n_rows();
const uword n_cols = P.get_n_cols();
const uword n_elem = P.get_n_elem();
out.set_size(1, n_elem);
eT* outmem = out.memptr();
if(n_cols == 1)
{
if(is_Mat<typename Proxy<T1>::stored_type>::value)
{
const unwrap<typename Proxy<T1>::stored_type> tmp(P.Q);
arrayops::copy(out.memptr(), tmp.M.memptr(), n_elem);
}
else
{
for(uword i=0; i < n_elem; ++i) { outmem[i] = P.at(i,0); }
}
}
else
{
for(uword row=0; row < n_rows; ++row)
{
uword i,j;
for(i=0, j=1; j < n_cols; i+=2, j+=2)
{
const eT tmp_i = P.at(row,i);
const eT tmp_j = P.at(row,j);
*outmem = tmp_i; outmem++;
*outmem = tmp_j; outmem++;
}
if(i < n_cols)
{
*outmem = P.at(row,i); outmem++;
}
}
}
}
template<typename T1>
inline
void
op_vectorise_all::apply(Mat<typename T1::elem_type>& out, const Op<T1,op_vectorise_all>& in)
{
arma_debug_sigprint();
const uword dim = in.aux_uword_a;
if(dim == 0)
{
op_vectorise_col::apply_direct(out, in.m);
}
else
{
op_vectorise_row::apply_direct(out, in.m);
}
}
//
template<typename T1>
inline
void
op_vectorise_cube_col::apply(Mat<typename T1::elem_type>& out, const CubeToMatOp<T1, op_vectorise_cube_col>& in)
{
arma_debug_sigprint();
typedef typename T1::elem_type eT;
if(is_same_type< T1, subview_cube<eT> >::yes)
{
op_vectorise_cube_col::apply_subview(out, reinterpret_cast< const subview_cube<eT>& >(in.m));
}
else
{
if((is_Cube<T1>::value) || (is_Cube<typename ProxyCube<T1>::stored_type>::value) || (arma_config::openmp && ProxyCube<T1>::use_mp))
{
op_vectorise_cube_col::apply_unwrap(out, in.m);
}
else
{
op_vectorise_cube_col::apply_proxy(out, in.m);
}
}
}
template<typename eT>
inline
void
op_vectorise_cube_col::apply_subview(Mat<eT>& out, const subview_cube<eT>& sv)
{
arma_debug_sigprint();
const uword sv_nr = sv.n_rows;
const uword sv_nc = sv.n_cols;
const uword sv_ns = sv.n_slices;
out.set_size(sv.n_elem, 1);
eT* out_mem = out.memptr();
for(uword s=0; s < sv_ns; ++s)
for(uword c=0; c < sv_nc; ++c)
{
arrayops::copy(out_mem, sv.slice_colptr(s,c), sv_nr);
out_mem += sv_nr;
}
}
template<typename T1>
inline
void
op_vectorise_cube_col::apply_unwrap(Mat<typename T1::elem_type>& out, const T1& expr)
{
arma_debug_sigprint();
const unwrap_cube<T1> U(expr);
out.set_size(U.M.n_elem, 1);
arrayops::copy(out.memptr(), U.M.memptr(), U.M.n_elem);
}
template<typename T1>
inline
void
op_vectorise_cube_col::apply_proxy(Mat<typename T1::elem_type>& out, const T1& expr)
{
arma_debug_sigprint();
typedef typename T1::elem_type eT;
const ProxyCube<T1> P(expr);
const uword N = P.get_n_elem();
out.set_size(N, 1);
eT* outmem = out.memptr();
if(ProxyCube<T1>::use_at == false)
{
typename ProxyCube<T1>::ea_type A = P.get_ea();
uword i,j;
for(i=0, j=1; j < N; i+=2, j+=2)
{
const eT tmp_i = A[i];
const eT tmp_j = A[j];
outmem[i] = tmp_i;
outmem[j] = tmp_j;
}
if(i < N)
{
outmem[i] = A[i];
}
}
else
{
const uword nr = P.get_n_rows();
const uword nc = P.get_n_cols();
const uword ns = P.get_n_slices();
for(uword s=0; s < ns; ++s)
for(uword c=0; c < nc; ++c)
for(uword r=0; r < nr; ++r)
{
*outmem = P.at(r,c,s);
outmem++;
}
}
}
//! @}