diff --git a/misc/test/src/Python/test_FaustPy.py b/misc/test/src/Python/test_FaustPy.py
index 038f9428422d5df2f1dad364d9c90d8b495b3bf1..570800fa0bea2548655abe5fdde0640c3b5924de 100644
--- a/misc/test/src/Python/test_FaustPy.py
+++ b/misc/test/src/Python/test_FaustPy.py
@@ -1191,6 +1191,113 @@ class TestFaustFactory(unittest.TestCase):
self.assertLessEqual(count_nonzero(Mp[:,i]), k)
self.assertAlmostEqual(norm(Mp), 1)
+ def test_splincol(self):
+ from pyfaust.factparams import splincol
+ from random import randint
+ from numpy.random import rand
+ from numpy import count_nonzero
+ from numpy.linalg import norm
+ min_n, min_m = 5, 5
+ m = randint(min_m, 128)
+ n = randint(min_n, 128)
+ M = rand(m,n)
+ k = randint(1,m)
+ p = splincol((m,n),k)
+ Mp = p(M)
+ # TODO: define sparsity assertions to verify on Mp
+ self.assertAlmostEqual(norm(Mp), 1)
+
+ def test_sp(self):
+ from pyfaust.factparams import sp
+ from random import randint
+ from numpy.random import rand
+ from numpy import count_nonzero
+ from numpy.linalg import norm
+ min_n, min_m = 5, 5
+ m = randint(min_m, 128)
+ n = randint(min_n, 128)
+ M = rand(m,n)
+ k = randint(1,m*n)
+ p = sp((m,n),k)
+ Mp = p(M)
+ for i in range(0,n):
+ # np.savez('M.npz', M, Mp, k)
+ self.assertLessEqual(count_nonzero(Mp[:,i]), k)
+ self.assertAlmostEqual(norm(Mp), 1)
+
+ def test_supp(self):
+ from pyfaust.factparams import supp
+ from numpy.random import rand
+ from numpy.random import randn, permutation as randperm
+ from numpy.linalg import norm
+ from random import randint
+ min_n, min_m = 5, 5
+ m = randint(min_m, 128)
+ n = randint(min_n, 128)
+ M = rand(m,n)
+ k = randint(1, min(m,n))
+ nnz_rinds = randperm(m)[:k]
+ nnz_cinds = randperm(n)[:k]
+ S = np.zeros((m,n))
+ S[nnz_rinds, nnz_cinds] = 1
+ p = supp(S)
+ pM = p(M)
+ # same nnz number
+ self.assertEqual(np.count_nonzero(pM), np.count_nonzero(S))
+ # same support
+ self.assertTrue(np.allclose(pM != 0, S != 0))
+ # pM normalized (according to fro-norm)
+ self.assertAlmostEqual(norm(pM), 1)
+ # same projection without normalization
+ p = supp(S, normalized=False)
+ pM = p(M)
+ self.assertTrue(np.allclose(pM[pM != 0], M[S != 0]))
+
+ def test_const(self):
+ from pyfaust.factparams import const
+ from numpy.random import rand
+ from random import randint
+ min_n, min_m = 5, 5
+ m = randint(min_m, 128)
+ n = randint(min_n, 128)
+ M = rand(m,n)
+ C = rand(m,n)
+ p = const(C, normalized=False)
+ pM = p(M)
+ self.assertTrue(np.allclose(C, pM))
+
+ def test_normcol(self):
+ from pyfaust.factparams import normcol
+ from numpy.random import rand
+ from numpy.random import randn, permutation as randperm
+ from numpy.linalg import norm
+ from random import randint, random
+ min_n, min_m = 5, 5
+ m = randint(min_m, 128)
+ n = randint(min_n, 128)
+ M = rand(m,n)*random()*50
+ k = random()*50
+ p = normcol(M.shape,k, normalized=False)
+ pM = p(M)
+ for i in range(pM.shape[1]):
+ self.assertAlmostEqual(norm(pM[:,i]), k)
+
+ def test_normlin(self):
+ from pyfaust.factparams import normlin
+ from numpy.random import rand
+ from numpy.random import randn, permutation as randperm
+ from numpy.linalg import norm
+ from random import randint, random
+ min_n, min_m = 5, 5
+ m = randint(min_m, 128)
+ n = randint(min_n, 128)
+ M = rand(m,n)*random()*50
+ k = random()*50
+ p = normlin(M.shape,k, normalized=False)
+ pM = p(M)
+ for i in range(pM.shape[0]):
+ self.assertAlmostEqual(norm(pM[i,:]), k)
+
if __name__ == "__main__":
if(len(sys.argv)> 1):
# argv[1] is for adding a directory in PYTHONPATH
diff --git a/src/algorithm/factorization/faust_GivensFGFT.h b/src/algorithm/factorization/faust_GivensFGFT.h
index 51077efce28fd42c3268a56fa48fc164170f45b5..255100f3243241e557a1f8387ccf67d15af09826 100644
--- a/src/algorithm/factorization/faust_GivensFGFT.h
+++ b/src/algorithm/factorization/faust_GivensFGFT.h
@@ -21,6 +21,8 @@ namespace Faust {
* \brief This class implements the Givens FGFT algorithm.
* This algorithm is based on the classical Jacobi eigenvalues algorithm.
*
+ * See parent class GivensFGFTGen for documentation about members.
+ *
* References:
*
* [1] Le Magoarou L., Gribonval R. and Tremblay N., "Approximate fast
@@ -34,73 +36,13 @@ namespace Faust {
Faust::MatDense<FPP,DEVICE> C;
/** \brief Column vector for the rowwise minimization of C (i.e. maximization of L). */
Faust::Vect<FPP,DEVICE> C_min_row;
- /** \brief Pivot candidates q coordinates. */
-// int* q_candidates; /* default IndexType for underlying eigen matrix is int. */
protected:
- /** \brief The number of targeted transform factors (when only one Givens rotation is stored per factor).*/
-// int J;
- /** \brief Fourier matrix factorization matrices (Givens matrix). */
-// vector<Faust::MatSparse<FPP,DEVICE>> facts;
- /** \brief Diagonalization approximate of Laplacian. */
-// Faust::Vect<FPP,DEVICE> D;
- /** \brief Queue of errors (cf. calc_err()). */
-// vector<FPP2> errs;
- /** \brief Pivot choices (p, q) coordinates. */
-// vector<pair<int,int>> coord_choices;
-// FPP Lap_squared_fro_norm;
- /** \brief L iteration factor: L_i = S^T L_{i-1} S, initialized from Lap (with S being facts[i]). */
-// Faust::MatGeneric<FPP, DEVICE> *L;
/** \brief Rotation angle theta for the current iteration's Givens matrix. */
FPP2 theta;
- /* Precomputed model identity matrix to init. facts[ite] before update.
- * Identity matrix is completed later with cos/sin coefficients (in update_fact()).
- */
-
- /** \brief Defines the rows of facts[ite]. */
-// vector<int> fact_mod_row_ids;
-// /** \brief Defines the columns of facts[ite]. */
-// vector<int> fact_mod_col_ids;
-// /** \brief Defines the coefficients of facts[ite]. */
-// vector<FPP> fact_mod_values;
-
-
- /** \brief Ordered indices of D to get increasing eigenvalues along the diagonal. */
-// vector<int> ord_indices;
-
- /** \brief inverse permutation of ord_indices (needed to retrieve start undefined order). */
-// vector<int> inv_ord_indices;
- /** \brief True is the last fact (of facts) has been permuted */
-// bool last_fact_permuted;
- /** \brief Cache for the ordered D. */
-// Faust::Vect<FPP,DEVICE> ordered_D;
- /** \brief true if D has already been ordered (order_D() was called). */
-// bool is_D_ordered;
-// int D_order_dir;
-
- /** \brief The level of verbosity (0 for nothing, 1 for iteration numbers,...) */
-// unsigned int verbosity;
-
- /**
- * \brief Row index for the selected pivot in L.
- */
-// int p,
-// /**
-// * \brief Column index for the selected pivot in L.
-// */q;
-
- /**
- * \brief Current iteration number (and index to the current factor to update).
- */
-// unsigned int ite;
-
/** \brief In calc_theta() two values are calculated for theta, this boolean is set to true to always choose theta2 (useful for GivensFGFTParallel). */
bool always_theta2;
-// bool stoppingCritIsError;
-// double stoppingError;
-// bool errIsRel;
-
/**
* Function pointer to any step of the algorithm (internal purpose only).
*/
@@ -128,15 +70,6 @@ namespace Faust {
*/
virtual void next_step();
- /**
- * \brief Algo. main loop (facts.size() iterations).
- *
- * It's the entry point to user code.
- * It calls next_step() in order to execute one iteration.
- *
- */
-// void compute_facts();
-
/**
* \brief Algo. step 2.4
*
@@ -145,29 +78,12 @@ namespace Faust {
virtual void update_L();
protected:
- /** \brief Algo. step 2.1.
- */
virtual void choose_pivot();
- /** \brief Algo. step 2.1.1
- *
- * Computes the max of L or sorts it.
- */
virtual void max_L();
- /** \brief Algo. step 2.2.
- *
- * Computes theta angle for the current Givens factor.
- *
- */
void calc_theta();
- /**
- * \brief Algo. step 2.3.
- *
- * Updates the current Givens factor according to the pivot chosen for the iteration.
- *
- */
virtual void update_fact();
virtual void update_L(MatDense<FPP,Cpu> &);
@@ -188,134 +104,14 @@ namespace Faust {
void update_L_first(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
void update_L_second(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
- /**
- * \brief Algo. step 2.5.
- *
- * Updates the diagonal approximate D from L (the diagonal part is taken).
- *
- */
-// void update_D();
- /**
- * \brief Algo. step 2.6.
- *
- * Computes the error of approximation for the current iteration.
- *
- */
void update_err();
- /**
- * Order (sort) D ascendantly into ordered_D and keeps ordered indices in ord_indices.
- */
-// void order_D();
-
- /**
- * Order D into ascendantly or descendantly into order_D and keeps ordered indices in ord_indices.
- *
- * \param order -1 to order descendantly, 1 to order ascendantly.
- */
-// void order_D(int order);
-
-
public:
- /**
- * Returns the ordered indices of D to get increasing eigenvalues along the diagonal.
- *
- * @see order_D()
- */
-// const vector<int>& get_ord_indices();
-
-
- /**
- * Returns the vector of errors computed by calc_err() during the algorithm iterations.
- */
-// const vector<FPP2>& get_errs() const;
- /**
- * Returns the specific j-th iteration's error (computed in calc_err()).
- */
-// FPP2 get_err(int j) const;
-
- /**
- * Returns the diag. vector D in its current status (which is updated at each iteration).
- */
-// const Faust::Vect<FPP,DEVICE>& get_D(const bool ord=false);
-
- /**
- * \param ord 0 (no sort), 1 (ascendant sort), -1 (descendant sort).
- */
-// const Faust::Vect<FPP,DEVICE>& get_D(const int ord=0);
-
-
- /** \brief Returns the diagonal vector as a sparse matrix.
- *
- * \note This function makes copies, is not intented for repeated use (use get_D() for optimized calls).
- *
- **/
const Faust::MatSparse<FPP,DEVICE> get_Dspm(const bool ord=false);
- /**
- * Returns the diagonal by copying it in the buffer diag_data (should be allocated by the callee).
- *
- * \param ord true to get the data ordered by ascendant eigenvalues false otherwise.
- */
-// void get_D(FPP* diag_data, const bool ord=false);
-
- /**
- *
- * \param ord: 0 for no sort, -1 for descending order, 1 for descending order.
- */
-// void get_D(FPP* diag_data, const int ord=0);
-
-
- /**
- * Computes and returns the Fourier matrix approximate from the Givens factors computed up to this time.
- *
- */
-// const MatDense<FPP,DEVICE> compute_fourier(const bool ord=false);
-
- /**
- * Returns the matrix L.
- *
- * @note this matrix is not the Laplacian matrix from which the algorithm has started from. This is a matrix initialized to the Laplacian before the first iteration and then is computed such that L = S^t L. S being the current Givens factor matrix updated (i.e. facts[ite]).
- *
- * @see get_Lap()
- *
- */
- const MatGeneric<FPP,DEVICE>& get_L() const ;
-
- /**
- * Returns the vector of all pivot choices (p, q) coordinates made by the algorithm until the last iteration.
- */
-// const vector<pair<int,int>>& get_coord_choices() const;
-
- /**
- * Returns the j-th iteration's pivot choice (p, q).
- */
-// void get_coord_choice(int j, int& p, int& q) const;
-
- /**
- * Returns the Laplacian matrix unmodified (as it was when passed to the constructor).
- */
-// const MatDense<FPP,DEVICE>& get_Lap() const;
-
- /**
- * Returns the vector of Givens matrices at this stage of algorithm execution (terminated or not).
- */
-// const vector<MatSparse<FPP,DEVICE>>& get_facts() const;
-
-
- /**
- * Returns a Faust::Transform object with copy of facts into it.
- *
- * \param ord true to get the Transform's facts ordering the last one columns according to ascendant eigenvalues, false to let facts as they went out from the algorithm (without reordering).
- */
-// Faust::Transform<FPP,DEVICE> get_transform(bool ord);
- /**
- * \param ord -1 for descending order, 1 for ascending order.
- */
-// Faust::Transform<FPP,DEVICE> get_transform(int ord);
};
diff --git a/src/algorithm/factorization/faust_GivensFGFT.hpp b/src/algorithm/factorization/faust_GivensFGFT.hpp
index b5cf23f6cd55117ba5080dfd519e67cc0ca58a6e..ec920ad6609413c3cb2553d3204f2e5f8f0568a0 100644
--- a/src/algorithm/factorization/faust_GivensFGFT.hpp
+++ b/src/algorithm/factorization/faust_GivensFGFT.hpp
@@ -1,12 +1,5 @@
using namespace Faust;
-// handle Visual Studio's whim https://msdn.microsoft.com/en-us/library/4hwaceh6.aspx
-#define _USE_MATH_DEFINES
-#include <cmath>
-#ifndef M_PI_4
-#define M_PI_4 (M_PI/4.0)
-#endif
-
template<typename FPP, Device DEVICE, typename FPP2>
void GivensFGFT<FPP,DEVICE,FPP2>::next_step()
{
@@ -348,18 +341,6 @@ void GivensFGFT<FPP,DEVICE,FPP2>::update_L_first(Eigen::SparseMatrix<FPP, RowMaj
L.mat.innerVector(q) = tmp;
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFT<FPP,DEVICE,FPP2>::update_D()
-//{
-// // D = spdiag(diag(L))
-// for(int i=0;i<L->getNbRow();i++)
-// D.getData()[i] = (*L)(i,i);
-//#ifdef DEBUG_GIVENS
-// D.Display();
-// cout << "D fro. norm: " << D.norm() << endl;
-//#endif
-//}
-
template<typename FPP, Device DEVICE, typename FPP2>
void GivensFGFT<FPP,DEVICE,FPP2>::update_err()
{
@@ -397,299 +378,27 @@ void GivensFGFT<FPP,DEVICE,FPP2>::update_err()
}
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFT<FPP,DEVICE,FPP2>::order_D()
-//{
-// order_D(1);
-//}
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFT<FPP,DEVICE,FPP2>::order_D(const int order /* -1 for descending order, 1 for ascending order */)
-//{
-// ordered_D = Faust::Vect<FPP,DEVICE>(D.size());
-// ord_indices.resize(0);
-// for(int i=0;i<D.size();i++)
-// ord_indices.push_back(i);
-// sort(ord_indices.begin(), ord_indices.end(), [this, &order](int i, int j) {
-// return order>0?D.getData()[i] < D.getData()[j]:(order <0?D.getData()[i] > D.getData()[j]:0);
-// });
-// for(int i=0;i<ord_indices.size();i++)
-// ordered_D.getData()[i] = D.getData()[ord_indices[i]];
-// // compute inverse permutation to keep easily possible retrieving undefined order
-// inv_ord_indices.resize(ord_indices.size());
-// int j = 0, i = 0;
-// while(j<ord_indices.size())
-// if(ord_indices[i] == j)
-// {
-// inv_ord_indices[j++] = i;
-// i = 0;
-// }
-// else
-// i++;
-// is_D_ordered = true;
-// D_order_dir = order;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<int>& GivensFGFT<FPP,DEVICE,FPP2>::get_ord_indices()
-//{
-// if(! is_D_ordered)
-// order_D();
-// return ord_indices;
-//}
-
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFT<FPP,DEVICE,FPP2>::compute_facts()
-//{
-// is_D_ordered = false; // facts (re)computed then D must be reordered
-// ite = 0;
-// bool stopping = false;
-// while(J == 0 || ite < facts.size()) // when J == 0 the stopping criterion is the error against Lap
-// {
-// next_step();
-// ite++;
-// if(stopping = (ite > 1 && stoppingCritIsError && errs.size() > 2 && errs[ite-1]-errs[ite-2] > FLT_EPSILON))
-// if(verbosity>0) cerr << "warning: the algorithm stopped because the last error is greater than the previous one." << endl;
-// if(stopping || stoppingCritIsError && errs.size() > 0 && (*(errs.end()-1) - stoppingError ) < FLT_EPSILON)
-// {
-// facts.resize(ite);
-// break;
-// }
-// }
-//}
-//
template<typename FPP, Device DEVICE, typename FPP2>
-GivensFGFT<FPP,DEVICE,FPP2>::GivensFGFT(Faust::MatSparse<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError /* default to 0.0 */, const bool errIsRel) : GivensFGFTGen<FPP,DEVICE,FPP2>(Lap, J, verbosity, stoppingError, errIsRel), /* Lap(Lap), facts(J>0?J:1), D(Lap.getNbRow()), */ C(Lap.getNbRow(), Lap.getNbCol()), /* errs(0), coord_choices(0), q_candidates(new int[Lap.getNbCol()]), is_D_ordered(false),*/ always_theta2(false) /*,verbosity(verbosity), stoppingCritIsError(stoppingError != 0.0), stoppingError(stoppingError), errIsRel(errIsRel), last_fact_permuted(false), Lap_squared_fro_norm(0), J(J)*/
-
+GivensFGFT<FPP,DEVICE,FPP2>::GivensFGFT(Faust::MatSparse<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError /* default to 0.0 */, const bool errIsRel) : GivensFGFTGen<FPP,DEVICE,FPP2>(Lap, J, verbosity, stoppingError, errIsRel), C(Lap.getNbRow(), Lap.getNbCol()), always_theta2(false)
{
- /* Matlab ref. code:
- * facts = cell(1,J);
- * n=size(Lap,1);
- * L=Lap;
- * C = 15*ones(n);
- * err=zeros(1,J);
- * coord_choices = zeros(2,J);
- *
- */
-// if(J == 0 && ! stoppingCritIsError) handleError("GivensFGFT", "Either J or stoppingError must be > 0");
+ //see parent ctor
C.setOnes();
C.scalarMultiply(15); // purely abitrary
-// if(Lap.getNbCol() != Lap.getNbRow())
-// handleError("Faust::GivensFGFT", "Laplacian must be a square matrix.");
-//
-// // init the identity part of the factor buffer model
-// // allocate the mem. space for the 4 additional rotation part coeffs
-// for(int i=0;i<Lap.getNbRow();i++)
-// {
-// fact_mod_values.push_back(FPP(1));
-// fact_mod_col_ids.push_back(i);
-// fact_mod_row_ids.push_back(i);
-// }
-//
-// // init. D
-// memset(D.getData(), 0, sizeof(FPP)*Lap.getNbRow());
-//
-// L = new MatSparse<FPP,DEVICE>(Lap);
}
template<typename FPP, Device DEVICE, typename FPP2>
-GivensFGFT<FPP,DEVICE,FPP2>::GivensFGFT(Faust::MatDense<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError, const bool errIsRel) : GivensFGFTGen<FPP,DEVICE,FPP2>(Lap, J, verbosity, stoppingError, errIsRel), /* Lap(Lap), facts(J>0?J:1), D(Lap.getNbRow()), */ C(Lap.getNbRow(), Lap.getNbCol()),/* errs(0), coord_choices(0), q_candidates(new int[Lap.getNbCol()]), is_D_ordered(false),*/ always_theta2(false) /*, verbosity(verbosity), stoppingCritIsError(stoppingError != 0.0), stoppingError(stoppingError), errIsRel(errIsRel), last_fact_permuted(false), Lap_squared_fro_norm(0), J(J) */
+GivensFGFT<FPP,DEVICE,FPP2>::GivensFGFT(Faust::MatDense<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError, const bool errIsRel) : GivensFGFTGen<FPP,DEVICE,FPP2>(Lap, J, verbosity, stoppingError, errIsRel), C(Lap.getNbRow(), Lap.getNbCol()), always_theta2(false)
{
- /* Matlab ref. code:
- * facts = cell(1,J);
- * n=size(Lap,1);
- * L=Lap;
- * C = 15*ones(n);
- * err=zeros(1,J);
- * coord_choices = zeros(2,J);
- *
- */
-// if(J == 0 && ! stoppingCritIsError) handleError("GivensFGFT", "Either J or stoppingError must be > 0");
+ // see parent ctor
C.setOnes();
C.scalarMultiply(15); // purely abitrary
-// if(Lap.getNbCol() != Lap.getNbRow())
-// handleError("Faust::GivensFGFT", "Laplacian must be a square matrix.");
-//
-// // init the identity part of the factor buffer model
-// // allocate the mem. space for the 4 additional rotation part coeffs
-// for(int i=0;i<Lap.getNbRow();i++)
-// {
-// fact_mod_values.push_back(FPP(1));
-// fact_mod_col_ids.push_back(i);
-// fact_mod_row_ids.push_back(i);
-// }
-//
-// // init. D
-// memset(D.getData(), 0, sizeof(FPP)*Lap.getNbRow());
-//
-// L = new MatDense<FPP,DEVICE>(Lap);
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//FPP2 GivensFGFT<FPP,DEVICE,FPP2>::get_err(int j) const
-//{
-// if(j > 0 && j < errs.size())
-// return errs[j];
-// else
-// throw out_of_range("GivensFGFT::get_err(j): j is out of range.");
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<FPP2>& GivensFGFT<FPP,DEVICE,FPP2>::get_errs() const
-//{
-// return errs;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::Vect<FPP,DEVICE>& GivensFGFT<FPP,DEVICE,FPP2>::get_D(const bool ord /* default to false */)
-//{
-// if(ord)
-// {
-// if(!is_D_ordered)
-// order_D();
-// return ordered_D;
-// }
-// return D;
-//}
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::Vect<FPP,DEVICE>& GivensFGFT<FPP,DEVICE,FPP2>::get_D(const int ord /* default to false */)
-//{
-// if(ord != 0)
-// {
-// if(!is_D_ordered || ord != D_order_dir)
-// order_D(ord);
-// return ordered_D;
-// }
-// return D;
-//}
-
template<typename FPP, Device DEVICE, typename FPP2>
const Faust::MatSparse<FPP,DEVICE> GivensFGFT<FPP,DEVICE,FPP2>::get_Dspm(const bool ord /* default to false */)
{
MatSparse <FPP,DEVICE> spD;
GivensFGFTGen<FPP,DEVICE,FPP2>::get_Dspm(spD, ord);
return spD;
-// const Faust::Vect<FPP,DEVICE>& D_ = this->get_D(ord);
-// vector<int> nat_ord_indices;
-// vector<FPP> diag_D;
-// for(int i=0;i<D_.size();i++)
-// {
-// nat_ord_indices.push_back(i);
-// diag_D.push_back(D_.getData()[i]);
-// }
-// MatSparse<FPP,DEVICE> spD(nat_ord_indices, nat_ord_indices, diag_D, nat_ord_indices.size(), nat_ord_indices.size());
-// return spD;
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFT<FPP,DEVICE,FPP2>::get_D(FPP* diag_data, const bool ord /* default to false */)
-//{
-// get_D(diag_data, ord?1:0);
-//}
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFT<FPP,DEVICE,FPP2>::get_D(FPP* diag_data, const int ord /* default to false */)
-//{
-// const Faust::Vect<FPP,DEVICE>& D_ = get_D(ord);
-// const FPP* src_data_ptr = D_.getData();
-// memcpy(diag_data, src_data_ptr, sizeof(FPP)*D_.size());
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::MatDense<FPP,DEVICE> GivensFGFT<FPP,DEVICE,FPP2>::compute_fourier(const bool ord /* default to false */)
-//{
-// Faust::MatDense<FPP,Cpu> fourier(Lap.getNbRow(), Lap.getNbCol());
-// Faust::MatDense<FPP,Cpu>* ord_fourier;
-// fourier.setEyes();
-// for(int i=facts.size()-1;i>=0;i--)
-// facts[i].multiply(fourier, 'N');
-// if(ord)
-// {
-// if(!is_D_ordered)
-// order_D();
-// ord_fourier = fourier.get_cols(ord_indices);
-// fourier = *ord_fourier;
-// delete ord_fourier;
-// }
-// return fourier;
-//}
-
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::MatGeneric<FPP,DEVICE>& GivensFGFT<FPP,DEVICE,FPP2>::get_L() const
-//{
-// return *L;
-//}
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<pair<int,int>>& GivensFGFT<FPP,DEVICE,FPP2>::get_coord_choices() const
-//{
-// return coord_choices;
-//}
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFT<FPP,DEVICE,FPP2>::get_coord_choice(int j, int& p, int& q) const
-//{
-// if(j > 0 && j < coord_choices.size())
-// {
-// p = coord_choices[j].first;
-// q = coord_choices[j].second;
-// }
-// else
-// throw out_of_range("GivensFGFT::get_coord_choice(j,p,q): j is out of range.");
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::MatDense<FPP,DEVICE>& GivensFGFT<FPP,DEVICE,FPP2>::get_Lap() const
-//{
-// return Lap;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<Faust::MatSparse<FPP,DEVICE>>& GivensFGFT<FPP,DEVICE,FPP2>::get_facts() const
-//{
-// return facts;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//Faust::Transform<FPP,DEVICE> GivensFGFT<FPP,DEVICE,FPP2>::get_transform(bool ord)
-//{
-// return get_transform(ord?1:0);
-//}
-//
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//Faust::Transform<FPP,DEVICE> GivensFGFT<FPP,DEVICE,FPP2>::get_transform(int ord)
-//{
-// //TODO: an optimization is possible by changing type of facts to vector<MatGeneric*> it would avoid copying facts into Transform and rather use them directly. It will need a destructor that deletes them eventually if they weren't transfered to a Transform object before.
-// MatSparse<FPP,DEVICE> & last_fact = *(facts.end()-1);
-// MatSparse<FPP,DEVICE> P(last_fact.getNbCol(), last_fact.getNbCol()); //last_fact permutation matrix
-// // (to reorder eigenvector transform according to ordered D)
-//
-// if(last_fact_permuted)
-// {
-// // get back to undefined order
-// for(int i=0;i<inv_ord_indices.size();i++)
-// P.setCoeff(inv_ord_indices[i],i, FPP(1.0));
-// last_fact.multiplyRight(P);
-// last_fact_permuted = false;
-// }
-//
-// if(ord)
-// {
-// if(!is_D_ordered || ord != D_order_dir)
-// order_D(ord);
-// for(int i=0;i<ord_indices.size();i++)
-// P.setCoeff(ord_indices[i],i, FPP(1.0));
-// // P.set_orthogonal(true);
-// // facts.push_back(P); // we prefer to directly multiply the last factor by P
-// last_fact_permuted = true;
-// last_fact.multiplyRight(P);
-// }
-// Faust::Transform<FPP,DEVICE> t = Faust::Transform<FPP,DEVICE>(facts);
-// // // remove the permutation factor if added temporarily for reordering
-// // return ord?facts.erase(facts.end()-1),t:t;
-// return t;
-//}
diff --git a/src/algorithm/factorization/faust_GivensFGFTComplex.h b/src/algorithm/factorization/faust_GivensFGFTComplex.h
index 13698030962264b31315e9805aa8317ba4d5e14e..3c8078a74397f1341048662e18ffbec2d4e6f17c 100644
--- a/src/algorithm/factorization/faust_GivensFGFTComplex.h
+++ b/src/algorithm/factorization/faust_GivensFGFTComplex.h
@@ -20,6 +20,8 @@ namespace Faust {
* \brief This class implements the Givens FGFT algorithm (Truncated Jacobi algorithm) for the complex matrix case (ideal case being the Hermitian matrix case).
* This algorithm is based on the classical Jacobi eigenvalues algorithm.
*
+ * See parent class GivensFGFTGen for documentation about members.
+ *
* References:
*
* [1] Le Magoarou L., Gribonval R. and Tremblay N., "Approximate fast
@@ -36,71 +38,11 @@ namespace Faust {
Faust::MatDense<FPP,DEVICE> C;
/** \brief Column vector for the rowwise minimization of C (i.e. maximization of L). */
Faust::Vect<FPP,DEVICE> C_min_row;
- /** \brief Pivot candidates q coordinates. */
- //int* q_candidates; /* default IndexType for underlying eigen matrix is int. */
public:
using DType = typename FPP::value_type;
protected:
- /** \brief The number of targeted transform factors (when only one Givens rotation is stored per factor).*/
-// int J;
- /** \brief Fourier matrix/eigenvectors factorization matrices (Givens matrix). */
-// vector<Faust::MatSparse<FPP,DEVICE>> facts;
- /** \brief Diagonalization approximate of Laplacian. */
- //Faust::Vect<DType,DEVICE> D;
- /** \brief Queue of errors (cf. calc_err()). */
-// vector<FPP2> errs;
- /** \brief Pivot choices (p, q) coordinates. */
-// vector<pair<int,int>> coord_choices;
- /** \brief Graph Laplacian to diagonalize/approximate. */
-// Faust::MatGeneric<FPP, DEVICE>& Lap;
-// FPP2 Lap_squared_fro_norm;
- /** \brief L iteration factor: L_i = S^T L_{i-1} S, initialized from Lap (with S being facts[i]). */
-// Faust::MatGeneric<FPP, DEVICE> *L;
- /** \brief Rotation angle theta for the current iteration's Givens matrix. */
+ /** \brief Rotation angle for the current iteration's "Givens" matrix. */
FPP theta1, theta2;
-// bool last_fact_permuted;
- /* Precomputed model identity matrix to init. facts[ite] before update.
- * Identity matrix is completed later with cos/sin coefficients (in update_fact()).
- */
-
- /** \brief Defines the rows of facts[ite]. */
-// vector<int> fact_mod_row_ids;
-// /** \brief Defines the columns of facts[ite]. */
-// vector<int> fact_mod_col_ids;
-// /** \brief Defines the coefficients of facts[ite]. */
-// vector<FPP> fact_mod_values;
-
-
- /** \brief Ordered indices of D to get increasing eigenvalues along the diagonal. */
-// vector<int> ord_indices;
- /** \brief inverse permutation of ord_indices (needed to retrieve start undefined order). */
-// vector<int> inv_ord_indices;
-
- /** \brief Cache for the ordered D. */
-// Faust::Vect<DType,DEVICE> ordered_D;
- /** \brief true if D has already been ordered (order_D() was called). */
- //bool is_D_ordered;
- //int D_order_dir;
-
- /** \brief The level of verbosity (0 for nothing, 1 for iteration numbers,...) */
- //unsigned int verbosity;
-
- /**
- * \brief Row index for the selected pivot in L.
- */
-// int p,
-// /**
-// * \brief Column index for the selected pivot in L.
-// */q;
-//
- /**
- * \brief Current iteration number (and index to the current factor to update).
- */
- // unsigned int ite;
-
-// bool stoppingCritIsError;
-// double stoppingError;
-// bool errIsRel;
/**
* Function pointer to any step of the algorithm (internal purpose only).
@@ -111,56 +53,18 @@ namespace Faust {
public:
const static unsigned int ERROR_CALC_PERIOD = 100;
- /** Algorithm class constructor.
- * \param Lap The Laplacian matrix to approximate/diagonalize.
- * \param J The number of iterations, Givens rotations factors.
- * */
GivensFGFTComplex(Faust::MatSparse<FPP,DEVICE>& Lap, int J, unsigned int verbosity = 0, const double stoppingError = 0.0, const bool errIsRel = true);
GivensFGFTComplex(Faust::MatDense<FPP,DEVICE>& Lap, int J, unsigned int verbosity = 0, const double stoppingError = 0.0, const bool errIsRel = true);
- /** Destructor */
-// virtual ~GivensFGFTComplex() {delete L;};
- /**
- * \brief Algo. main step.
- *
- * This function calls all step functions of the algorithm in the proper order to execute only one iteration.
- *
- * External code may call this function instead of compute_facts() in order to debug iterations taken one by one.
- */
virtual void next_step();
- /**
- * \brief Algo. main loop (facts.size() iterations).
- *
- * It's the entry point to user code.
- * It calls next_step() in order to execute one iteration.
- *
- */
-// void compute_facts();
-
- /**
- * \brief Algo. step 2.4
- *
- * Updates L after Givens factor update for the next iteration.
- */
virtual void update_L();
protected:
- /** \brief Algo. step 2.1.
- */
virtual void choose_pivot();
- /** \brief Algo. step 2.1.1
- *
- * Computes the max of L or sorts it.
- */
virtual void max_L();
- /** \brief Algo. step 2.2.
- *
- * Computes theta angle for the current Givens factor.
- *
- */
void calc_theta();
/**
@@ -168,12 +72,6 @@ namespace Faust {
*/
void check_pivot_image(FPP& c_pp, FPP& c_pq, FPP& c_qp, FPP& c_qq);
- /**
- * \brief Algo. step 2.3.
- *
- * Updates the current Givens factor according to the pivot chosen for the iteration.
- *
- */
virtual void update_fact();
virtual void update_L(MatDense<FPP,Cpu> &);
@@ -194,136 +92,13 @@ namespace Faust {
void update_L_first(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP& c_pp, const FPP& c_pq, const FPP& c_qp, const FPP& c_qq, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
void update_L_second(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP& c_pp, const FPP& c_pq, const FPP& c_qp, const FPP& c_qq, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
- /**
- * \brief Algo. step 2.5.
- *
- * Updates the diagonal approximate D from L (the diagonal part is taken).
- *
- */
-// void update_D();
- /**
- * \brief Algo. step 2.6.
- *
- * Computes the error of approximation for the current iteration.
- *
- */
void update_err();
- /**
- * Order (sort) D ascendantly into ordered_D and keeps ordered indices in ord_indices.
- */
-// void order_D();
-
- /**
- * Order D into ascendantly or descendantly into order_D and keeps ordered indices in ord_indices.
- *
- * \param order -1 to order descendantly, 1 to order ascendantly.
- */
-// void order_D(int order);
-
-
public:
- /**
- * Returns the ordered indices of D to get increasing eigenvalues along the diagonal.
- *
- * @see order_D()
- */
-// const vector<int>& get_ord_indices();
-
-
- /**
- * Returns the vector of errors computed by calc_err() during the algorithm iterations.
- */
-// const vector<FPP2>& get_errs() const;
-
- /**
- * Returns the specific j-th iteration's error (computed in calc_err()).
- */
-// FPP2 get_err(int j) const;
-
- /**
- * Returns the diag. vector D in its current status (which is updated at each iteration).
- */
-// const Faust::Vect<DType /*FPP*/,DEVICE>& get_D(const bool ord=false);
-
- /**
- * \param ord 0 (no sort), 1 (ascendant sort), -1 (descendant sort).
- */
-// const Faust::Vect<DType /*FPP*/,DEVICE>& get_D(const int ord=0);
-
-
- /** \brief Returns the diagonal vector as a sparse matrix.
- *
- * \note This function makes copies, is not intented for repeated use (use get_D() for optimized calls).
- *
- **/
const Faust::MatSparse<FPP,DEVICE> get_Dspm(const bool ord=false);
- /**
- * Returns the diagonal by copying it in the buffer diag_data (should be allocated by the callee).
- *
- * \param ord true to get the data ordered by ascendant eigenvalues false otherwise.
- */
-// void get_D(DType/*FPP*/* diag_data, const bool ord=false);
-
- /**
- *
- * \param ord: 0 for no sort, -1 for descending order, 1 for descending order.
- */
-// void get_D(DType/*FPP*/* diag_data, const int ord=0);
-
-
- /**
- * Computes and returns the Fourier matrix approximate from the Givens factors computed up to this time.
- *
- */
-// const MatDense<FPP,DEVICE> compute_fourier(const bool ord=false);
-
- /**
- * Returns the matrix L.
- *
- * @note this matrix is not the Laplacian matrix from which the algorithm has started from. This is a matrix initialized to the Laplacian before the first iteration and then is computed such that L = S^t L. S being the current Givens factor matrix updated (i.e. facts[ite]).
- *
- * @see get_Lap()
- *
- */
-// const MatGeneric<FPP,DEVICE>& get_L() const ;
-
- /**
- * Returns the vector of all pivot choices (p, q) coordinates made by the algorithm until the last iteration.
- */
-// const vector<pair<int,int>>& get_coord_choices() const;
-
- /**
- * Returns the j-th iteration's pivot choice (p, q).
- */
-// void get_coord_choice(int j, int& p, int& q) const;
-
- /**
- * Returns the Laplacian matrix unmodified (as it was when passed to the constructor).
- */
-// const MatDense<FPP,DEVICE>& get_Lap() const;
-
- /**
- * Returns the vector of Givens matrices at this stage of algorithm execution (terminated or not).
- */
-// const vector<MatSparse<FPP,DEVICE>>& get_facts() const;
-
-
- /**
- * Returns a Faust::Transform object with copy of facts into it.
- *
- * \param ord true to get the Transform's facts ordering the last one columns according to ascendant eigenvalues, false to let facts as they went out from the algorithm (without reordering).
- */
-// Faust::Transform<FPP,DEVICE> get_transform(bool ord);
- /**
- * \param ord -1 for descending order, 1 for ascending order.
- */
-// Faust::Transform<FPP,DEVICE> get_transform(int ord);
-
-
};
}
diff --git a/src/algorithm/factorization/faust_GivensFGFTComplex.hpp b/src/algorithm/factorization/faust_GivensFGFTComplex.hpp
index 796ea66b208b9985adcb6fc36d36e54b67c5a062..b44cb3719df12a36b4f384a3fb101e3bb22656f1 100644
--- a/src/algorithm/factorization/faust_GivensFGFTComplex.hpp
+++ b/src/algorithm/factorization/faust_GivensFGFTComplex.hpp
@@ -1,12 +1,5 @@
using namespace Faust;
-// handle Visual Studio's whim https://msdn.microsoft.com/en-us/library/4hwaceh6.aspx
-#define _USE_MATH_DEFINES
-#include <cmath>
-#ifndef M_PI_4
-#define M_PI_4 (M_PI/4.0)
-#endif
-
template<typename FPP, Device DEVICE, typename FPP2>
void GivensFGFTComplex<FPP,DEVICE,FPP2>::next_step()
{
@@ -414,12 +407,6 @@ void GivensFGFTComplex<FPP,DEVICE,FPP2>::update_L_first(Eigen::SparseMatrix<FPP,
L.mat.innerVector(q) = tmp;
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFTComplex<FPP,DEVICE,FPP2>::update_D()
-//{
-// GivensFGFTGen<typename FPP::value_type, DEVICE, FPP2, FPP>::update_D(this->L);
-//}
-
template<typename FPP, Device DEVICE, typename FPP2>
void GivensFGFTComplex<FPP,DEVICE,FPP2>::update_err()
{
@@ -457,303 +444,26 @@ void GivensFGFTComplex<FPP,DEVICE,FPP2>::update_err()
}
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFTComplex<FPP,DEVICE,FPP2>::order_D()
-//{
-// order_D(1);
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFTComplex<FPP,DEVICE,FPP2>::order_D(const int order /* -1 for descending order, 1 for ascending order */)
-//{
-// ordered_D = Faust::Vect<DType /*FPP*/,DEVICE>(this->D.size());
-// ord_indices.resize(0);
-// for(int i=0;i<this->D.size();i++)
-// ord_indices.push_back(i);
-// sort(ord_indices.begin(), ord_indices.end(), [this, &order](int i, int j) {
-// return order>0?this->D.getData()[i]/*.real()*/ < this->D.getData()[j]/*.real()*/:(order <0?this->D.getData()[i]/*.real()*/ > this->D.getData()[j]/*.real()*/:0);
-// });
-// for(int i=0;i<ord_indices.size();i++)
-// {
-// ordered_D.getData()[i] = this->D.getData()[ord_indices[i]];
-// }
-// // compute inverse permutation to keep easily possible retrieving undefined order
-// inv_ord_indices.resize(ord_indices.size());
-// int j = 0, i = 0;
-// while(j<ord_indices.size())
-// if(ord_indices[i] == j)
-// {
-// inv_ord_indices[j++] = i;
-// i = 0;
-// }
-// else
-// i++;
-// is_D_ordered = true;
-// D_order_dir = order;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<int>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_ord_indices()
-//{
-// if(! is_D_ordered)
-// order_D();
-// return ord_indices;
-//}
-
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFTComplex<FPP,DEVICE,FPP2>::compute_facts()
-//{
-// this->is_D_ordered = false; // facts (re)computed then D must be reordered
-// this->ite = 0;
-// bool stopping = false;
-// while(this->J == 0 || this->ite < this->facts.size()) // when J == 0 the stopping criterion is the error against Lap
-// {
-// next_step();
-// this->ite++;
-// if(stopping = (this->ite > 1 && this->stoppingCritIsError && this->errs.size() > 2 && this->errs[this->ite-1]-this->errs[this->ite-2] > FLT_EPSILON))
-// if(this->verbosity>0) cerr << "warning: the algorithm stopped because the last error is greater than the previous one." << endl;
-// if(stopping || this->stoppingCritIsError && this->errs.size() > 0 && (*(this->errs.end()-1) - this->stoppingError ) < FLT_EPSILON)
-// {
-// this->facts.resize(this->ite);
-// break;
-// }
-// }
-//}
-
template<typename FPP, Device DEVICE, typename FPP2>
-GivensFGFTComplex<FPP,DEVICE,FPP2>::GivensFGFTComplex(Faust::MatSparse<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError /* default to 0.0 */, const bool errIsRel) : Faust::GivensFGFTGen<typename FPP::value_type, DEVICE, FPP2, FPP>(Lap, J, verbosity, stoppingError, errIsRel)/*, Lap(Lap), facts(J>0?J:1), D(Lap.getNbRow())*/, C(Lap.getNbRow(), Lap.getNbCol())/*,errs(0), coord_choices(0), q_candidates(new int[Lap.getNbCol()]), is_D_ordered(false), verbosity(verbosity), stoppingCritIsError(stoppingError != 0.0), stoppingError(stoppingError), errIsRel(errIsRel),last_fact_permuted(false), Lap_squared_fro_norm(FPP2(0)) J(J)*/
+GivensFGFTComplex<FPP,DEVICE,FPP2>::GivensFGFTComplex(Faust::MatSparse<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError /* default to 0.0 */, const bool errIsRel) : Faust::GivensFGFTGen<typename FPP::value_type, DEVICE, FPP2, FPP>(Lap, J, verbosity, stoppingError, errIsRel), C(Lap.getNbRow(), Lap.getNbCol())
{
- /* Matlab ref. code:
- * facts = cell(1,J);
- * n=size(Lap,1);
- * L=Lap;
- * C = 15*ones(n);
- * err=zeros(1,J);
- * coord_choices = zeros(2,J);
- *
- */
-// if(this->J == 0 && ! this->stoppingCritIsError) handleError("GivensFGFT", "Either J or stoppingError must be > 0");
+ //see parent ctor
this->C.setZeros();
- // C.setOnes();
- // C.scalarMultiply(-15); // purely abitrary
-// if(Lap.getNbCol() != Lap.getNbRow())
-// handleError("Faust::GivensFGFTComplex", "Laplacian must be a square matrix.");
-
-// // init the identity part of the factor buffer model
-// // allocate the mem. space for the 4 additional rotation part coeffs
-// for(int i=0;i<Lap.getNbRow();i++)
-// {
-// fact_mod_values.push_back(FPP(1));
-// fact_mod_col_ids.push_back(i);
-// fact_mod_row_ids.push_back(i);
-// }
-
- // init. D
-// memset(this->D.getData(), 0, sizeof(DType/*FPP*/)*Lap.getNbRow());
-
-// L = new MatSparse<FPP,DEVICE>(Lap);
}
template<typename FPP, Device DEVICE, typename FPP2>
-GivensFGFTComplex<FPP,DEVICE,FPP2>::GivensFGFTComplex(Faust::MatDense<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError, const bool errIsRel) : Faust::GivensFGFTGen<typename FPP::value_type, DEVICE, FPP2, FPP>(Lap, J, verbosity, stoppingError, errIsRel) /* Lap(Lap), facts(J>0?J:1), D(Lap.getNbRow())*/, C(Lap.getNbRow(), Lap.getNbCol())/*, errs(0), coord_choices(0), q_candidates(new int[Lap.getNbCol()]), is_D_ordered(false), verbosity(verbosity), stoppingCritIsError(stoppingError != 0.0), stoppingError(stoppingError), errIsRel(errIsRel), last_fact_permuted(false), Lap_squared_fro_norm(FPP2(0)) J(J)*/
+GivensFGFTComplex<FPP,DEVICE,FPP2>::GivensFGFTComplex(Faust::MatDense<FPP,DEVICE>& Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError, const bool errIsRel) : Faust::GivensFGFTGen<typename FPP::value_type, DEVICE, FPP2, FPP>(Lap, J, verbosity, stoppingError, errIsRel), C(Lap.getNbRow(), Lap.getNbCol())
{
- /* Matlab ref. code:
- * facts = cell(1,J);
- * n=size(Lap,1);
- * L=Lap;
- * C = 15*ones(n);
- * err=zeros(1,J);
- * coord_choices = zeros(2,J);
- *
- */
-// if(this->J == 0 && ! this->stoppingCritIsError) handleError("GivensFGFT", "Either J or stoppingError must be > 0");
+ // see parent ctor
this->C.setZeros();
-// C.setOnes();
-// C.scalarMultiply(FPP(-15)); // purely abitrary
-// if(Lap.getNbCol() != Lap.getNbRow())
-// handleError("Faust::GivensFGFTComplex", "Laplacian must be a square matrix.");
-
-// // init the identity part of the factor buffer model
-// // allocate the mem. space for the 4 additional rotation part coeffs
-// for(int i=0;i<Lap.getNbRow();i++)
-// {
-// fact_mod_values.push_back(FPP(1));
-// fact_mod_col_ids.push_back(i);
-// fact_mod_row_ids.push_back(i);
-// }
-
- // init. D
- //memset(this->D.getData(), 0, sizeof(DType/*FPP*/)*Lap.getNbRow());
-
-// L = new MatDense<FPP,DEVICE>(Lap);
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//FPP2 GivensFGFTComplex<FPP,DEVICE,FPP2>::get_err(int j) const
-//{
-// if(j > 0 && j < errs.size())
-// return errs[j];
-// else
-// throw out_of_range("GivensFGFTComplex::get_err(j): j is out of range.");
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<FPP2>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_errs() const
-//{
-// return errs;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::Vect<typename FPP::value_type/*FPP*/,DEVICE>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_D(const bool ord /* default to false */)
-//{
-// if(ord)
-// {
-// if(!is_D_ordered)
-// order_D();
-// return ordered_D;
-// }
-// return D;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::Vect<typename FPP::value_type/*FPP*/,DEVICE>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_D(const int ord /* default to false */)
-//{
-// if(ord != 0)
-// {
-// if(!is_D_ordered || ord != D_order_dir)
-// order_D(ord);
-// return ordered_D;
-// }
-// return D;
-//}
-
template<typename FPP, Device DEVICE, typename FPP2>
-const Faust::MatSparse<FPP,DEVICE> GivensFGFTComplex<FPP,DEVICE,FPP2>::get_Dspm(const bool ord /* default to false*/)
+const Faust::MatSparse<FPP,DEVICE> GivensFGFTComplex<FPP,DEVICE,FPP2>::get_Dspm(const bool ord /* default to false*/)
{
MatSparse <FPP,DEVICE> spD;
GivensFGFTGen<typename FPP::value_type,DEVICE,FPP2,FPP>::get_Dspm(spD, ord);
return spD;
-// const Faust::Vect<FPP,DEVICE>& D_ = this->get_D(ord);
-// vector<int> nat_ord_indices;
-// vector<FPP> diag_D;
-// for(int i=0;i<D_.size();i++)
-// {
-// nat_ord_indices.push_back(i);
-// diag_D.push_back(FPP(D_.getData()[i]));
-// }
-// MatSparse<FPP,DEVICE> spD(nat_ord_indices, nat_ord_indices, diag_D, nat_ord_indices.size(), nat_ord_indices.size());
-// return spD;
}
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFTComplex<FPP,DEVICE,FPP2>::get_D(DType* /*FPP**/ diag_data, const bool ord /* default to false */)
-//{
-// get_D(diag_data, ord?1:0);
-//}
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFTComplex<FPP,DEVICE,FPP2>::get_D(DType* /*FPP**/ diag_data, const int ord /* default to false */)
-//{
-// const Faust::Vect<FPP,DEVICE>& D_ = get_D(ord);
-// const DType* /*FPP**/ src_data_ptr = D_.getData();
-// memcpy(diag_data, src_data_ptr, sizeof(FPP)*D_.size());
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::MatDense<FPP,DEVICE> GivensFGFTComplex<FPP,DEVICE,FPP2>::compute_fourier(const bool ord /* default to false */)
-//{
-// Faust::MatDense<FPP,Cpu> fourier(L->getNbRow(), L->getNbCol());
-// Faust::MatDense<FPP,Cpu>* ord_fourier;
-// fourier.setEyes();
-// for(int i=facts.size()-1;i>=0;i--)
-// facts[i].multiply(fourier, 'N');
-// if(ord)
-// {
-// if(!this->is_D_ordered)
-// this->order_D();
-// ord_fourier = fourier.get_cols(this->ord_indices);
-// fourier = *ord_fourier;
-// delete ord_fourier;
-// }
-// return fourier;
-//}
-
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::MatGeneric<FPP,DEVICE>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_L() const
-//{
-// return *L;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<pair<int,int>>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_coord_choices() const
-//{
-// return coord_choices;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//void GivensFGFTComplex<FPP,DEVICE,FPP2>::get_coord_choice(int j, int& p, int& q) const
-//{
-// if(j > 0 && j < coord_choices.size())
-// {
-// p = coord_choices[j].first;
-// q = coord_choices[j].second;
-// }
-// else
-// throw out_of_range("GivensFGFTComplex::get_coord_choice(j,p,q): j is out of range.");
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const Faust::MatDense<FPP,DEVICE>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_Lap() const
-//{
-// return Lap;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//const vector<Faust::MatSparse<FPP,DEVICE>>& GivensFGFTComplex<FPP,DEVICE,FPP2>::get_facts() const
-//{
-// return this->facts;
-//}
-
-//template<typename FPP, Device DEVICE, typename FPP2>
-//Faust::Transform<FPP,DEVICE> GivensFGFTComplex<FPP,DEVICE,FPP2>::get_transform(bool ord)
-//{
-// return get_transform(ord?1:0);
-//}
-//
-//
-//template<typename FPP, Device DEVICE, typename FPP2>
-//Faust::Transform<FPP,DEVICE> GivensFGFTComplex<FPP,DEVICE,FPP2>::get_transform(int ord)
-//{
-// //TODO: an optimization is possible by changing type of facts to vector<MatGeneric*> it would avoid copying facts into Transform and rather use them directly. It will need a destructor that deletes them eventually if they weren't transfered to a Transform object before.
-// MatSparse<FPP,DEVICE> & last_fact = *(this->facts.end()-1);
-// MatSparse<FPP,DEVICE> P(last_fact.getNbCol(), last_fact.getNbCol()); //last_fact permutation matrix
-// // (to reorder eigenvector transform according to ordered D)
-//
-// if(last_fact_permuted)
-// {
-// // non-ordered eigenvector transform (ord == 0) or opposite order asked (ord != D_order_dir)
-// // get back to undefined order
-// for(int i=0;i<this->inv_ord_indices.size();i++)
-// P.setCoeff(this->inv_ord_indices[i],i, FPP(1.0));
-// last_fact.multiplyRight(P);
-// last_fact_permuted = false;
-// }
-//
-// if(ord)
-// {
-// if(!this->is_D_ordered || ord != this->D_order_dir)
-// this->order_D(ord);
-// for(int i=0;i<this->ord_indices.size();i++)
-// P.setCoeff(this->ord_indices[i],i, FPP(1.0));
-// // P.set_orthogonal(true);
-// // facts.push_back(P); // we prefer to directly multiply the last factor by P
-// last_fact_permuted = true;
-// last_fact.multiplyRight(P);
-// }
-// Faust::Transform<FPP,DEVICE> t = Faust::Transform<FPP,DEVICE>(this->facts);
-// // // remove the permutation factor if added temporarily for reordering
-// // return ord?facts.erase(facts.end()-1),t:t;
-// return t;
-//}
+
diff --git a/src/algorithm/factorization/faust_GivensFGFTGen.h b/src/algorithm/factorization/faust_GivensFGFTGen.h
index b6ca12ae0adcd9efc9c4360b83506ba118a96689..7aa756da7fcf92cb0fc7b6b09cd88e3d81052a8c 100644
--- a/src/algorithm/factorization/faust_GivensFGFTGen.h
+++ b/src/algorithm/factorization/faust_GivensFGFTGen.h
@@ -30,11 +30,11 @@ namespace Faust {
*
*/
/** \brief Temporary storage matrix for maximization of L. */
-// Faust::MatDense<FPP4,DEVICE> C;
+ // Faust::MatDense<FPP4,DEVICE> C;
/** \brief Column vector for the rowwise minimization of C (i.e. maximization of L). */
-// Faust::Vect<FPP4,DEVICE> C_min_row;
+ // Faust::Vect<FPP4,DEVICE> C_min_row;
protected:
- /** \brief Fourier matrix/eigenvectors factorization matrices (Givens matrix). */
+ /** \brief Fourier matrix/eigenvectors factorization matrices (Givens matrices). */
vector<Faust::MatSparse<FPP4,DEVICE>> facts;
/** \brief L iteration factor: L_i = S^T L_{i-1} S, initialized from Lap (with S being facts[i]). */
@@ -44,22 +44,20 @@ namespace Faust {
/** \brief The number of targeted transform factors (when only one Givens rotation is stored per factor).*/
int J;
- /** \brief Fourier matrix factorization matrices (Givens matrix). */
-// vector<Faust::MatSparse<FPP,DEVICE>> facts;
- /** \brief Diagonalization approximate of Laplacian. */
+ /** \brief approximate eigenvalues vector. */
Faust::Vect<FPP,DEVICE> D;
- /** \brief Queue of errors (cf. calc_err()). */
+ /** \brief Queue of errors (cf. update_err()). */
vector<FPP2> errs;
/** \brief Pivot choices (p, q) coordinates. */
vector<pair<int,int>> coord_choices;
/** \brief Graph Laplacian to diagonalize/approximate. */
Faust::MatGeneric<FPP4, DEVICE>& Lap;
+ /** \brief Laplacian dimension */
unsigned int dim_size;
+ /** \brief Laplacian Frobenius norm */
FPP2 Lap_squared_fro_norm;
- /** \brief L iteration factor: L_i = S^T L_{i-1} S, initialized from Lap (with S being facts[i]). */
-// Faust::MatGeneric<FPP, DEVICE> *L;
/** \brief Rotation angle theta for the current iteration's Givens matrix. */
-// FPP2 theta;
+ // FPP2 theta;
/* Precomputed model identity matrix to init. facts[ite] before update.
* Identity matrix is completed later with cos/sin coefficients (in update_fact()).
@@ -84,6 +82,7 @@ namespace Faust {
Faust::Vect<FPP,DEVICE> ordered_D;
/** \brief true if D has already been ordered (order_D() was called). */
bool is_D_ordered;
+ /** \brief 1 if eigenvalues has been ordered in ascending order -1 ortherwise (other values is for undefined order). */
int D_order_dir;
/** \brief The level of verbosity (0 for nothing, 1 for iteration numbers,...) */
@@ -102,33 +101,27 @@ namespace Faust {
*/
unsigned int ite;
- /** \brief In calc_theta() two values are calculated for theta, this boolean is set to true to always choose theta2 (useful for GivensFGFTGenParallel). */
-// bool always_theta2;
-
+ /** \brief true if the stopping criterion is error (otherwise it's the number of iterations/number of Givens matrices) */
bool stoppingCritIsError;
+ /** \brief error value according to algorithm stops if stoppingCritIsError == true */
double stoppingError;
+ /** \brief true if the stopping error is taken as relative error (absolute otherwise). */
bool errIsRel;
- /**
- * Function pointer to any step of the algorithm (internal purpose only).
- */
-// typedef void (GivensFGFTGen<FPP,DEVICE,FPP2>::*substep_fun)();
-
public:
-// const static unsigned int ERROR_CALC_PERIOD = 100;
+ // const static unsigned int ERROR_CALC_PERIOD = 100;
/** Algorithm class constructor.
* \param Lap The Laplacian matrix to approximate/diagonalize.
* \param J The number of iterations, Givens rotations factors.
+ * TODO: complete argument list
* */
GivensFGFTGen(Faust::MatGeneric<FPP4,DEVICE>* Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError, const bool errIsRel);
GivensFGFTGen(Faust::MatSparse<FPP4, DEVICE> & Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError, const bool errIsRel);
GivensFGFTGen(Faust::MatDense<FPP4, DEVICE> & Lap, int J, unsigned int verbosity /* deft val == 0 */, const double stoppingError, const bool errIsRel);
-// GivensFGFTGen(Faust::MatSparse<FPP,DEVICE>& Lap, int J, unsigned int verbosity = 0, const double stoppingError = 0.0, const bool errIsRel = true);
-// GivensFGFTGen(Faust::MatDense<FPP,DEVICE>& Lap, int J, unsigned int verbosity = 0, const double stoppingError = 0.0, const bool errIsRel = true);
/** Destructor */
virtual ~GivensFGFTGen() {delete[] q_candidates; delete L;};
@@ -183,24 +176,24 @@ namespace Faust {
*/
virtual void update_fact()=0;
-// virtual void update_L(MatDense<FPP,Cpu> &);
-//
-// virtual void update_L(MatSparse<FPP,Cpu> &);
+ // virtual void update_L(MatDense<FPP,Cpu> &);
+ //
+ // virtual void update_L(MatSparse<FPP,Cpu> &);
/**
* Computes the first S'*L (only used by update_L() in optimization enabled code).
*
*/
-// void update_L_first(Faust::Vect<FPP,DEVICE>& L_vec_p, Faust::Vect<FPP,DEVICE>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, MatDense<FPP,DEVICE> & L);
+ // void update_L_first(Faust::Vect<FPP,DEVICE>& L_vec_p, Faust::Vect<FPP,DEVICE>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, MatDense<FPP,DEVICE> & L);
/**
* Computes L*S (only used by update_L() in optimization enabled code).
* Must be called after update_L_first() to finish the update of L = S'*L*S.
*/
-// void update_L_second(Faust::Vect<FPP,DEVICE>& L_vec_p, Faust::Vect<FPP,DEVICE>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, MatDense<FPP,DEVICE> & L);
+ // void update_L_second(Faust::Vect<FPP,DEVICE>& L_vec_p, Faust::Vect<FPP,DEVICE>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, MatDense<FPP,DEVICE> & L);
-// void update_L_first(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
-// void update_L_second(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
+ // void update_L_first(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
+ // void update_L_second(Eigen::SparseMatrix<FPP, RowMajor> & L_vec_p, Eigen::SparseMatrix<FPP, RowMajor>& L_vec_q, const FPP2& c, const FPP2& s, int p, int q, Faust::MatSparse<FPP,DEVICE> & L);
/**
* \brief Algo. step 2.5.
*
@@ -218,14 +211,14 @@ namespace Faust {
virtual void update_err()=0;
/**
- * Order (sort) D ascendantly into ordered_D and keeps ordered indices in ord_indices.
+ * Sort D in descending order into ordered_D and keeps ordered indices in ord_indices.
*/
void order_D();
/**
- * Order D into ascendantly or descendantly into order_D and keeps ordered indices in ord_indices.
+ * Sort D into order_D and keeps ordered indices in ord_indices.
*
- * \param order -1 to order descendantly, 1 to order ascendantly.
+ * \param order -1 for descending order, 1 to ascending order.
*/
void order_D(int order);
@@ -260,15 +253,13 @@ namespace Faust {
*/
const Faust::Vect<FPP,DEVICE>& get_D(const int ord=0);
- template<typename FPP3>
- void get_Dspm(Faust::MatSparse<FPP3,DEVICE>&, const bool ord=false);
-
/** \brief Returns the diagonal vector as a sparse matrix.
*
* \note This function makes copies, is not intented for repeated use (use get_D() for optimized calls).
*
**/
-// const Faust::MatSparse<FPP,DEVICE> get_Dspm(const bool ord=false);
+ template<typename FPP3>
+ void get_Dspm(Faust::MatSparse<FPP3,DEVICE>&, const bool ord=false);
/**
* Returns the diagonal by copying it in the buffer diag_data (should be allocated by the callee).
diff --git a/wrapper/python/pyfaust/factparams.py b/wrapper/python/pyfaust/factparams.py
index c7fcf61eeb83a67132b506221e0b29d31e36a5a4..2528ae80f9c5a8335f0cb39707336da9baf9cdf2 100644
--- a/wrapper/python/pyfaust/factparams.py
+++ b/wrapper/python/pyfaust/factparams.py
@@ -285,7 +285,8 @@ class ConstraintMat(ConstraintGeneric):
np.ndarray)):
raise TypeError('ConstraintMat must receive a numpy matrix as cons_value '
'argument.')
- self.cons_value = self._cons_value
+ self.cons_value = np.asfortranarray(self._cons_value)
+ self._cons_value = self.cons_value
if(normalized == None):
if(self.name.name == ContraintName.CONST):
# for const proj the default is to not normalize
@@ -335,7 +336,7 @@ class ConstraintReal(ConstraintGeneric):
>>> from numpy.linalg import norm
>>> cons = ConstraintReal('normcol', 10, 10, 2.) # a short for ConstraintReal(ConstraintName(ConstraintName.NORMCOL), 10, 10, 2.)
>>> M = rand(10,10)*10
- >>> norm(M[:,2],2)
+ >>> norm(M[:,2])
17.041462424512272
>>> norm(cons.project(M)[:,2])
2.0
diff --git a/wrapper/python/src/_FaustCorePy.pyx b/wrapper/python/src/_FaustCorePy.pyx
index c7e36fe8afa4a72f8fe673962eb94a3f00c6e75e..4e4e3667d78999cf1905234925b0ff9e5567f040 100644
--- a/wrapper/python/src/_FaustCorePy.pyx
+++ b/wrapper/python/src/_FaustCorePy.pyx
@@ -858,10 +858,9 @@ cdef class ConstraintMatCore:
'float16', 'float32',
'float64', 'double']
- order = 'C'
- if(np.isfortran(M)):
- order = 'FORTRAN'
- M_out = np.empty(M.shape, dtype=M.dtype, order=order)
+ M = np.asfortranarray(M)
+
+ M_out = np.empty(M.shape, dtype=M.dtype, order='F')
check_matrix(isReal, M)
@@ -902,12 +901,11 @@ cdef class ConstraintRealCore:
'float16', 'float32',
'float64', 'double']
-
-
- order = 'C'
- if(np.isfortran(M)):
- order = 'FORTRAN'
- M_out = np.empty(M.shape, dtype=M.dtype, order=order)
+# order = 'C'
+# if(np.isfortran(M)):
+# order = 'FORTRAN'
+ M = np.asfortranarray(M)
+ M_out = np.empty(M.shape, dtype=M.dtype, order='F')
check_matrix(isReal, M)
if(isReal):