Unfinished version of Armadillo-using AD class added.

This commit is contained in:
Atgeirr Flø Rasmussen 2013-05-06 09:43:26 +02:00
parent 0a43fe68dd
commit e59e5cc5f2

216
AutoDiffBlockArma.hpp Normal file
View File

@ -0,0 +1,216 @@
/*
Copyright 2013 SINTEF ICT, Applied Mathematics.
This file is part of the Open Porous Media project (OPM).
OPM 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.
OPM 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 OPM. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef OPM_AUTODIFFBLOCKARMA_HEADER_INCLUDED
#define OPM_AUTODIFFBLOCKARMA_HEADER_INCLUDED
// #include "AutoDiff.hpp"
// #include <Eigen/Eigen>
// #include <Eigen/Sparse>
#include <armadillo>
#include <vector>
#include <cassert>
namespace AutoDiff
{
template <typename Scalar>
class ForwardBlock
{
public:
/// Underlying types for scalar vectors and jacobians.
typedef arma::Col<Scalar> V;
typedef arma::SpMat<Scalar> M;
/// Named constructor pattern used here.
static ForwardBlock constant(const V& val, const std::vector<int>& blocksizes)
{
std::vector<M> jac;
const int num_elem = val.size();
const int num_blocks = blocksizes.size();
// For constants, all jacobian blocks are zero.
jac.resize(num_blocks);
for (int i = 0; i < num_blocks; ++i) {
jac[i] = M(num_elem, blocksizes[i]);
}
return ForwardBlock(val, jac);
}
static ForwardBlock variable(const int index, const V& val, const std::vector<int>& blocksizes)
{
std::vector<M> jac;
const int num_elem = val.size();
const int num_blocks = blocksizes.size();
// First, set all jacobian blocks to zero...
jac.resize(num_blocks);
for (int i = 0; i < num_blocks; ++i) {
jac[i] = M(num_elem, blocksizes[i]);
}
// ... then set the one corrresponding to this variable to identity.
assert(blocksizes[index] == num_elem);
jac[index].eye(num_elem, num_elem);
return ForwardBlock(val, jac);
}
static ForwardBlock function(const V& val, const std::vector<M>& jac)
{
return ForwardBlock(val, jac);
}
/// Operator +
ForwardBlock operator+(const ForwardBlock& rhs)
{
std::vector<M> jac = jac_;
assert(numBlocks() == rhs.numBlocks());
int num_blocks = numBlocks();
for (int block = 0; block < num_blocks; ++block) {
assert(jac[block].n_rows == rhs.jac_[block].n_rows);
assert(jac[block].n_cols == rhs.jac_[block].n_cols);
jac[block] += rhs.jac_[block];
}
return function(val_ + rhs.val_, jac);
}
/// Operator -
ForwardBlock operator-(const ForwardBlock& rhs)
{
std::vector<M> jac = jac_;
assert(numBlocks() == rhs.numBlocks());
int num_blocks = numBlocks();
for (int block = 0; block < num_blocks; ++block) {
assert(jac[block].n_rows == rhs.jac_[block].n_rows);
assert(jac[block].n_cols == rhs.jac_[block].n_cols);
jac[block] -= rhs.jac_[block];
}
return function(val_ - rhs.val_, jac);
}
/// Operator *
ForwardBlock operator*(const ForwardBlock& rhs)
{
int num_blocks = numBlocks();
std::vector<M> jac(num_blocks);
assert(numBlocks() == rhs.numBlocks());
typedef Eigen::DiagonalMatrix<Scalar, Eigen::Dynamic> D;
D D1 = val_.matrix().asDiagonal();
D D2 = rhs.val_.matrix().asDiagonal();
for (int block = 0; block < num_blocks; ++block) {
assert(jac_[block].n_rows == rhs.jac_[block].n_rows);
assert(jac_[block].n_cols == rhs.jac_[block].n_cols);
jac[block] = D2*jac_[block] + D1*rhs.jac_[block];
}
return function(val_ * rhs.val_, jac);
}
/// Operator /
ForwardBlock operator/(const ForwardBlock& rhs)
{
int num_blocks = numBlocks();
std::vector<M> jac(num_blocks);
assert(numBlocks() == rhs.numBlocks());
typedef Eigen::DiagonalMatrix<Scalar, Eigen::Dynamic> D;
D D1 = val_.matrix().asDiagonal();
D D2 = rhs.val_.matrix().asDiagonal();
D D3 = std::pow(rhs.val_, -2).matrix().asDiagonal();
for (int block = 0; block < num_blocks; ++block) {
assert(jac_[block].n_rows == rhs.jac_[block].n_rows);
assert(jac_[block].n_cols == rhs.jac_[block].n_cols);
jac[block] = D3 * (D2*jac_[block] - D1*rhs.jac_[block]);
}
return function(val_ / rhs.val_, jac);
}
/// I/O.
template <class Ostream>
Ostream&
print(Ostream& os) const
{
int num_blocks = jac_.size();
os << "Value =\n" << val_ << "\n\nJacobian =\n";
for (int i = 0; i < num_blocks; ++i) {
os << "Sub Jacobian #" << i << '\n' << jac_[i] << "\n";
}
return os;
}
/// Number of variables or Jacobian blocks.
int numBlocks() const
{
return jac_.size();
}
/// Function value
const V& value() const
{
return val_;
}
/// Function derivatives
const std::vector<M>& derivative() const
{
return jac_;
}
private:
ForwardBlock(const V& val,
const std::vector<M>& jac)
: val_(val), jac_(jac)
{
#ifndef NDEBUG
const int num_elem = val_.size();
const int num_blocks = jac_.size();
for (int block = 0; block < num_blocks; ++block) {
assert(num_elem == jac_[block].n_rows);
}
#endif
}
V val_;
std::vector<M> jac_;
};
template <class Ostream, typename Scalar>
Ostream&
operator<<(Ostream& os, const ForwardBlock<Scalar>& fw)
{
return fw.print(os);
}
/// Multiply with sparse matrix from the left.
template <typename Scalar>
ForwardBlock<Scalar> operator*(const typename ForwardBlock<Scalar>::M& lhs,
const ForwardBlock<Scalar>& rhs)
{
int num_blocks = rhs.numBlocks();
std::vector<typename ForwardBlock<Scalar>::M> jac(num_blocks);
assert(lhs.n_cols == rhs.value().n_rows);
for (int block = 0; block < num_blocks; ++block) {
jac[block] = lhs*rhs.derivative()[block];
}
typename ForwardBlock<Scalar>::V val = lhs*rhs.value().matrix();
return ForwardBlock<Scalar>::function(val, jac);
}
} // namespace Autodiff
#endif // OPM_AUTODIFFBLOCKARMA_HEADER_INCLUDED