NDDEM
JacobiSVD.h
Go to the documentation of this file.
1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 2009-2010 Benoit Jacob <jacob.benoit.1@gmail.com>
5 // Copyright (C) 2013-2014 Gael Guennebaud <gael.guennebaud@inria.fr>
6 //
7 // This Source Code Form is subject to the terms of the Mozilla
8 // Public License v. 2.0. If a copy of the MPL was not distributed
9 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
10 
11 #ifndef EIGEN_JACOBISVD_H
12 #define EIGEN_JACOBISVD_H
13 
14 namespace Eigen {
15 
16 namespace internal {
17 // forward declaration (needed by ICC)
18 // the empty body is required by MSVC
19 template<typename MatrixType, int QRPreconditioner,
22 
23 /*** QR preconditioners (R-SVD)
24  ***
25  *** Their role is to reduce the problem of computing the SVD to the case of a square matrix.
26  *** This approach, known as R-SVD, is an optimization for rectangular-enough matrices, and is a requirement for
27  *** JacobiSVD which by itself is only able to work on square matrices.
28  ***/
29 
31 
32 template<typename MatrixType, int QRPreconditioner, int Case>
34 {
35  enum { a = MatrixType::RowsAtCompileTime != Dynamic &&
36  MatrixType::ColsAtCompileTime != Dynamic &&
37  MatrixType::ColsAtCompileTime <= MatrixType::RowsAtCompileTime,
38  b = MatrixType::RowsAtCompileTime != Dynamic &&
39  MatrixType::ColsAtCompileTime != Dynamic &&
40  MatrixType::RowsAtCompileTime <= MatrixType::ColsAtCompileTime,
41  ret = !( (QRPreconditioner == NoQRPreconditioner) ||
42  (Case == PreconditionIfMoreColsThanRows && bool(a)) ||
43  (Case == PreconditionIfMoreRowsThanCols && bool(b)) )
44  };
45 };
46 
47 template<typename MatrixType, int QRPreconditioner, int Case,
50 
51 template<typename MatrixType, int QRPreconditioner, int Case>
52 class qr_preconditioner_impl<MatrixType, QRPreconditioner, Case, false>
53 {
54 public:
56  bool run(JacobiSVD<MatrixType, QRPreconditioner>&, const MatrixType&)
57  {
58  return false;
59  }
60 };
61 
62 /*** preconditioner using FullPivHouseholderQR ***/
63 
64 template<typename MatrixType>
66 {
67 public:
68  typedef typename MatrixType::Scalar Scalar;
69  enum
70  {
71  RowsAtCompileTime = MatrixType::RowsAtCompileTime,
72  MaxRowsAtCompileTime = MatrixType::MaxRowsAtCompileTime
73  };
75 
77  {
78  if (svd.rows() != m_qr.rows() || svd.cols() != m_qr.cols())
79  {
80  m_qr.~QRType();
81  ::new (&m_qr) QRType(svd.rows(), svd.cols());
82  }
83  if (svd.m_computeFullU) m_workspace.resize(svd.rows());
84  }
85 
87  {
88  if(matrix.rows() > matrix.cols())
89  {
90  m_qr.compute(matrix);
91  svd.m_workMatrix = m_qr.matrixQR().block(0,0,matrix.cols(),matrix.cols()).template triangularView<Upper>();
92  if(svd.m_computeFullU) m_qr.matrixQ().evalTo(svd.m_matrixU, m_workspace);
93  if(svd.computeV()) svd.m_matrixV = m_qr.colsPermutation();
94  return true;
95  }
96  return false;
97  }
98 private:
102 };
103 
104 template<typename MatrixType>
106 {
107 public:
108  typedef typename MatrixType::Scalar Scalar;
109  enum
110  {
111  RowsAtCompileTime = MatrixType::RowsAtCompileTime,
112  ColsAtCompileTime = MatrixType::ColsAtCompileTime,
113  MaxRowsAtCompileTime = MatrixType::MaxRowsAtCompileTime,
114  MaxColsAtCompileTime = MatrixType::MaxColsAtCompileTime,
115  Options = MatrixType::Options
116  };
117 
118  typedef typename internal::make_proper_matrix_type<
119  Scalar, ColsAtCompileTime, RowsAtCompileTime, Options, MaxColsAtCompileTime, MaxRowsAtCompileTime
121 
123  {
124  if (svd.cols() != m_qr.rows() || svd.rows() != m_qr.cols())
125  {
126  m_qr.~QRType();
127  ::new (&m_qr) QRType(svd.cols(), svd.rows());
128  }
129  m_adjoint.resize(svd.cols(), svd.rows());
130  if (svd.m_computeFullV) m_workspace.resize(svd.cols());
131  }
132 
134  {
135  if(matrix.cols() > matrix.rows())
136  {
137  m_adjoint = matrix.adjoint();
138  m_qr.compute(m_adjoint);
139  svd.m_workMatrix = m_qr.matrixQR().block(0,0,matrix.rows(),matrix.rows()).template triangularView<Upper>().adjoint();
140  if(svd.m_computeFullV) m_qr.matrixQ().evalTo(svd.m_matrixV, m_workspace);
141  if(svd.computeU()) svd.m_matrixU = m_qr.colsPermutation();
142  return true;
143  }
144  else return false;
145  }
146 private:
151 };
152 
153 /*** preconditioner using ColPivHouseholderQR ***/
154 
155 template<typename MatrixType>
157 {
158 public:
160  {
161  if (svd.rows() != m_qr.rows() || svd.cols() != m_qr.cols())
162  {
163  m_qr.~QRType();
164  ::new (&m_qr) QRType(svd.rows(), svd.cols());
165  }
166  if (svd.m_computeFullU) m_workspace.resize(svd.rows());
167  else if (svd.m_computeThinU) m_workspace.resize(svd.cols());
168  }
169 
171  {
172  if(matrix.rows() > matrix.cols())
173  {
174  m_qr.compute(matrix);
175  svd.m_workMatrix = m_qr.matrixQR().block(0,0,matrix.cols(),matrix.cols()).template triangularView<Upper>();
176  if(svd.m_computeFullU) m_qr.householderQ().evalTo(svd.m_matrixU, m_workspace);
177  else if(svd.m_computeThinU)
178  {
179  svd.m_matrixU.setIdentity(matrix.rows(), matrix.cols());
180  m_qr.householderQ().applyThisOnTheLeft(svd.m_matrixU, m_workspace);
181  }
182  if(svd.computeV()) svd.m_matrixV = m_qr.colsPermutation();
183  return true;
184  }
185  return false;
186  }
187 
188 private:
192 };
193 
194 template<typename MatrixType>
196 {
197 public:
198  typedef typename MatrixType::Scalar Scalar;
199  enum
200  {
201  RowsAtCompileTime = MatrixType::RowsAtCompileTime,
202  ColsAtCompileTime = MatrixType::ColsAtCompileTime,
203  MaxRowsAtCompileTime = MatrixType::MaxRowsAtCompileTime,
204  MaxColsAtCompileTime = MatrixType::MaxColsAtCompileTime,
205  Options = MatrixType::Options
206  };
207 
208  typedef typename internal::make_proper_matrix_type<
209  Scalar, ColsAtCompileTime, RowsAtCompileTime, Options, MaxColsAtCompileTime, MaxRowsAtCompileTime
211 
213  {
214  if (svd.cols() != m_qr.rows() || svd.rows() != m_qr.cols())
215  {
216  m_qr.~QRType();
217  ::new (&m_qr) QRType(svd.cols(), svd.rows());
218  }
219  if (svd.m_computeFullV) m_workspace.resize(svd.cols());
220  else if (svd.m_computeThinV) m_workspace.resize(svd.rows());
221  m_adjoint.resize(svd.cols(), svd.rows());
222  }
223 
225  {
226  if(matrix.cols() > matrix.rows())
227  {
228  m_adjoint = matrix.adjoint();
229  m_qr.compute(m_adjoint);
230 
231  svd.m_workMatrix = m_qr.matrixQR().block(0,0,matrix.rows(),matrix.rows()).template triangularView<Upper>().adjoint();
232  if(svd.m_computeFullV) m_qr.householderQ().evalTo(svd.m_matrixV, m_workspace);
233  else if(svd.m_computeThinV)
234  {
235  svd.m_matrixV.setIdentity(matrix.cols(), matrix.rows());
236  m_qr.householderQ().applyThisOnTheLeft(svd.m_matrixV, m_workspace);
237  }
238  if(svd.computeU()) svd.m_matrixU = m_qr.colsPermutation();
239  return true;
240  }
241  else return false;
242  }
243 
244 private:
249 };
250 
251 /*** preconditioner using HouseholderQR ***/
252 
253 template<typename MatrixType>
255 {
256 public:
258  {
259  if (svd.rows() != m_qr.rows() || svd.cols() != m_qr.cols())
260  {
261  m_qr.~QRType();
262  ::new (&m_qr) QRType(svd.rows(), svd.cols());
263  }
264  if (svd.m_computeFullU) m_workspace.resize(svd.rows());
265  else if (svd.m_computeThinU) m_workspace.resize(svd.cols());
266  }
267 
268  bool run(JacobiSVD<MatrixType, HouseholderQRPreconditioner>& svd, const MatrixType& matrix)
269  {
270  if(matrix.rows() > matrix.cols())
271  {
272  m_qr.compute(matrix);
273  svd.m_workMatrix = m_qr.matrixQR().block(0,0,matrix.cols(),matrix.cols()).template triangularView<Upper>();
274  if(svd.m_computeFullU) m_qr.householderQ().evalTo(svd.m_matrixU, m_workspace);
275  else if(svd.m_computeThinU)
276  {
277  svd.m_matrixU.setIdentity(matrix.rows(), matrix.cols());
278  m_qr.householderQ().applyThisOnTheLeft(svd.m_matrixU, m_workspace);
279  }
280  if(svd.computeV()) svd.m_matrixV.setIdentity(matrix.cols(), matrix.cols());
281  return true;
282  }
283  return false;
284  }
285 private:
289 };
290 
291 template<typename MatrixType>
293 {
294 public:
295  typedef typename MatrixType::Scalar Scalar;
296  enum
297  {
298  RowsAtCompileTime = MatrixType::RowsAtCompileTime,
299  ColsAtCompileTime = MatrixType::ColsAtCompileTime,
300  MaxRowsAtCompileTime = MatrixType::MaxRowsAtCompileTime,
301  MaxColsAtCompileTime = MatrixType::MaxColsAtCompileTime,
302  Options = MatrixType::Options
303  };
304 
305  typedef typename internal::make_proper_matrix_type<
306  Scalar, ColsAtCompileTime, RowsAtCompileTime, Options, MaxColsAtCompileTime, MaxRowsAtCompileTime
308 
310  {
311  if (svd.cols() != m_qr.rows() || svd.rows() != m_qr.cols())
312  {
313  m_qr.~QRType();
314  ::new (&m_qr) QRType(svd.cols(), svd.rows());
315  }
316  if (svd.m_computeFullV) m_workspace.resize(svd.cols());
317  else if (svd.m_computeThinV) m_workspace.resize(svd.rows());
318  m_adjoint.resize(svd.cols(), svd.rows());
319  }
320 
321  bool run(JacobiSVD<MatrixType, HouseholderQRPreconditioner>& svd, const MatrixType& matrix)
322  {
323  if(matrix.cols() > matrix.rows())
324  {
325  m_adjoint = matrix.adjoint();
326  m_qr.compute(m_adjoint);
327 
328  svd.m_workMatrix = m_qr.matrixQR().block(0,0,matrix.rows(),matrix.rows()).template triangularView<Upper>().adjoint();
329  if(svd.m_computeFullV) m_qr.householderQ().evalTo(svd.m_matrixV, m_workspace);
330  else if(svd.m_computeThinV)
331  {
332  svd.m_matrixV.setIdentity(matrix.cols(), matrix.rows());
333  m_qr.householderQ().applyThisOnTheLeft(svd.m_matrixV, m_workspace);
334  }
335  if(svd.computeU()) svd.m_matrixU.setIdentity(matrix.rows(), matrix.rows());
336  return true;
337  }
338  else return false;
339  }
340 
341 private:
346 };
347 
348 /*** 2x2 SVD implementation
349  ***
350  *** JacobiSVD consists in performing a series of 2x2 SVD subproblems
351  ***/
352 
353 template<typename MatrixType, int QRPreconditioner>
354 struct svd_precondition_2x2_block_to_be_real<MatrixType, QRPreconditioner, false>
355 {
357  typedef typename MatrixType::RealScalar RealScalar;
358  static bool run(typename SVD::WorkMatrixType&, SVD&, Index, Index, RealScalar&) { return true; }
359 };
360 
361 template<typename MatrixType, int QRPreconditioner>
362 struct svd_precondition_2x2_block_to_be_real<MatrixType, QRPreconditioner, true>
363 {
365  typedef typename MatrixType::Scalar Scalar;
366  typedef typename MatrixType::RealScalar RealScalar;
367  static bool run(typename SVD::WorkMatrixType& work_matrix, SVD& svd, Index p, Index q, RealScalar& maxDiagEntry)
368  {
369  using std::sqrt;
370  using std::abs;
373  RealScalar n = sqrt(numext::abs2(work_matrix.coeff(p,p)) + numext::abs2(work_matrix.coeff(q,p)));
374 
375  const RealScalar considerAsZero = (std::numeric_limits<RealScalar>::min)();
376  const RealScalar precision = NumTraits<Scalar>::epsilon();
377 
378  if(n==0)
379  {
380  // make sure first column is zero
381  work_matrix.coeffRef(p,p) = work_matrix.coeffRef(q,p) = Scalar(0);
382 
383  if(abs(numext::imag(work_matrix.coeff(p,q)))>considerAsZero)
384  {
385  // work_matrix.coeff(p,q) can be zero if work_matrix.coeff(q,p) is not zero but small enough to underflow when computing n
386  z = abs(work_matrix.coeff(p,q)) / work_matrix.coeff(p,q);
387  work_matrix.row(p) *= z;
388  if(svd.computeU()) svd.m_matrixU.col(p) *= conj(z);
389  }
390  if(abs(numext::imag(work_matrix.coeff(q,q)))>considerAsZero)
391  {
392  z = abs(work_matrix.coeff(q,q)) / work_matrix.coeff(q,q);
393  work_matrix.row(q) *= z;
394  if(svd.computeU()) svd.m_matrixU.col(q) *= conj(z);
395  }
396  // otherwise the second row is already zero, so we have nothing to do.
397  }
398  else
399  {
400  rot.c() = conj(work_matrix.coeff(p,p)) / n;
401  rot.s() = work_matrix.coeff(q,p) / n;
402  work_matrix.applyOnTheLeft(p,q,rot);
403  if(svd.computeU()) svd.m_matrixU.applyOnTheRight(p,q,rot.adjoint());
404  if(abs(numext::imag(work_matrix.coeff(p,q)))>considerAsZero)
405  {
406  z = abs(work_matrix.coeff(p,q)) / work_matrix.coeff(p,q);
407  work_matrix.col(q) *= z;
408  if(svd.computeV()) svd.m_matrixV.col(q) *= z;
409  }
410  if(abs(numext::imag(work_matrix.coeff(q,q)))>considerAsZero)
411  {
412  z = abs(work_matrix.coeff(q,q)) / work_matrix.coeff(q,q);
413  work_matrix.row(q) *= z;
414  if(svd.computeU()) svd.m_matrixU.col(q) *= conj(z);
415  }
416  }
417 
418  // update largest diagonal entry
419  maxDiagEntry = numext::maxi<RealScalar>(maxDiagEntry,numext::maxi<RealScalar>(abs(work_matrix.coeff(p,p)), abs(work_matrix.coeff(q,q))));
420  // and check whether the 2x2 block is already diagonal
421  RealScalar threshold = numext::maxi<RealScalar>(considerAsZero, precision * maxDiagEntry);
422  return abs(work_matrix.coeff(p,q))>threshold || abs(work_matrix.coeff(q,p)) > threshold;
423  }
424 };
425 
426 template<typename _MatrixType, int QRPreconditioner>
427 struct traits<JacobiSVD<_MatrixType,QRPreconditioner> >
428  : traits<_MatrixType>
429 {
430  typedef _MatrixType MatrixType;
431 };
432 
433 } // end namespace internal
434 
488 template<typename _MatrixType, int QRPreconditioner> class JacobiSVD
489  : public SVDBase<JacobiSVD<_MatrixType,QRPreconditioner> >
490 {
492  public:
493 
494  typedef _MatrixType MatrixType;
495  typedef typename MatrixType::Scalar Scalar;
497  enum {
498  RowsAtCompileTime = MatrixType::RowsAtCompileTime,
499  ColsAtCompileTime = MatrixType::ColsAtCompileTime,
501  MaxRowsAtCompileTime = MatrixType::MaxRowsAtCompileTime,
502  MaxColsAtCompileTime = MatrixType::MaxColsAtCompileTime,
504  MatrixOptions = MatrixType::Options
505  };
506 
507  typedef typename Base::MatrixUType MatrixUType;
508  typedef typename Base::MatrixVType MatrixVType;
510 
516 
523  {}
524 
525 
532  JacobiSVD(Index rows, Index cols, unsigned int computationOptions = 0)
533  {
534  allocate(rows, cols, computationOptions);
535  }
536 
547  explicit JacobiSVD(const MatrixType& matrix, unsigned int computationOptions = 0)
548  {
549  compute(matrix, computationOptions);
550  }
551 
562  JacobiSVD& compute(const MatrixType& matrix, unsigned int computationOptions);
563 
570  JacobiSVD& compute(const MatrixType& matrix)
571  {
572  return compute(matrix, m_computationOptions);
573  }
574 
575  using Base::computeU;
576  using Base::computeV;
577  using Base::rows;
578  using Base::cols;
579  using Base::rank;
580 
581  private:
582  void allocate(Index rows, Index cols, unsigned int computationOptions);
583 
584  protected:
585  using Base::m_matrixU;
586  using Base::m_matrixV;
588  using Base::m_info;
589  using Base::m_isInitialized;
590  using Base::m_isAllocated;
592  using Base::m_computeFullU;
593  using Base::m_computeThinU;
594  using Base::m_computeFullV;
595  using Base::m_computeThinV;
598  using Base::m_rows;
599  using Base::m_cols;
600  using Base::m_diagSize;
603 
604  template<typename __MatrixType, int _QRPreconditioner, bool _IsComplex>
606  template<typename __MatrixType, int _QRPreconditioner, int _Case, bool _DoAnything>
608 
612 };
613 
614 template<typename MatrixType, int QRPreconditioner>
615 void JacobiSVD<MatrixType, QRPreconditioner>::allocate(Eigen::Index rows, Eigen::Index cols, unsigned int computationOptions)
616 {
617  eigen_assert(rows >= 0 && cols >= 0);
618 
619  if (m_isAllocated &&
620  rows == m_rows &&
621  cols == m_cols &&
622  computationOptions == m_computationOptions)
623  {
624  return;
625  }
626 
627  m_rows = rows;
628  m_cols = cols;
629  m_info = Success;
630  m_isInitialized = false;
631  m_isAllocated = true;
632  m_computationOptions = computationOptions;
633  m_computeFullU = (computationOptions & ComputeFullU) != 0;
634  m_computeThinU = (computationOptions & ComputeThinU) != 0;
635  m_computeFullV = (computationOptions & ComputeFullV) != 0;
636  m_computeThinV = (computationOptions & ComputeThinV) != 0;
637  eigen_assert(!(m_computeFullU && m_computeThinU) && "JacobiSVD: you can't ask for both full and thin U");
638  eigen_assert(!(m_computeFullV && m_computeThinV) && "JacobiSVD: you can't ask for both full and thin V");
639  eigen_assert(EIGEN_IMPLIES(m_computeThinU || m_computeThinV, MatrixType::ColsAtCompileTime==Dynamic) &&
640  "JacobiSVD: thin U and V are only available when your matrix has a dynamic number of columns.");
641  if (QRPreconditioner == FullPivHouseholderQRPreconditioner)
642  {
643  eigen_assert(!(m_computeThinU || m_computeThinV) &&
644  "JacobiSVD: can't compute thin U or thin V with the FullPivHouseholderQR preconditioner. "
645  "Use the ColPivHouseholderQR preconditioner instead.");
646  }
647  m_diagSize = (std::min)(m_rows, m_cols);
648  m_singularValues.resize(m_diagSize);
649  if(RowsAtCompileTime==Dynamic)
650  m_matrixU.resize(m_rows, m_computeFullU ? m_rows
651  : m_computeThinU ? m_diagSize
652  : 0);
653  if(ColsAtCompileTime==Dynamic)
654  m_matrixV.resize(m_cols, m_computeFullV ? m_cols
655  : m_computeThinV ? m_diagSize
656  : 0);
657  m_workMatrix.resize(m_diagSize, m_diagSize);
658 
659  if(m_cols>m_rows) m_qr_precond_morecols.allocate(*this);
660  if(m_rows>m_cols) m_qr_precond_morerows.allocate(*this);
661  if(m_rows!=m_cols) m_scaledMatrix.resize(rows,cols);
662 }
663 
664 template<typename MatrixType, int QRPreconditioner>
666 JacobiSVD<MatrixType, QRPreconditioner>::compute(const MatrixType& matrix, unsigned int computationOptions)
667 {
668  using std::abs;
669  allocate(matrix.rows(), matrix.cols(), computationOptions);
670 
671  // currently we stop when we reach precision 2*epsilon as the last bit of precision can require an unreasonable number of iterations,
672  // only worsening the precision of U and V as we accumulate more rotations
673  const RealScalar precision = RealScalar(2) * NumTraits<Scalar>::epsilon();
674 
675  // limit for denormal numbers to be considered zero in order to avoid infinite loops (see bug 286)
676  const RealScalar considerAsZero = (std::numeric_limits<RealScalar>::min)();
677 
678  // Scaling factor to reduce over/under-flows
679  RealScalar scale = matrix.cwiseAbs().template maxCoeff<PropagateNaN>();
680  if (!(numext::isfinite)(scale)) {
681  m_isInitialized = true;
682  m_info = InvalidInput;
683  return *this;
684  }
685  if(scale==RealScalar(0)) scale = RealScalar(1);
686 
687  /*** step 1. The R-SVD step: we use a QR decomposition to reduce to the case of a square matrix */
688 
689  if(m_rows!=m_cols)
690  {
691  m_scaledMatrix = matrix / scale;
692  m_qr_precond_morecols.run(*this, m_scaledMatrix);
693  m_qr_precond_morerows.run(*this, m_scaledMatrix);
694  }
695  else
696  {
697  m_workMatrix = matrix.block(0,0,m_diagSize,m_diagSize) / scale;
698  if(m_computeFullU) m_matrixU.setIdentity(m_rows,m_rows);
699  if(m_computeThinU) m_matrixU.setIdentity(m_rows,m_diagSize);
700  if(m_computeFullV) m_matrixV.setIdentity(m_cols,m_cols);
701  if(m_computeThinV) m_matrixV.setIdentity(m_cols, m_diagSize);
702  }
703 
704  /*** step 2. The main Jacobi SVD iteration. ***/
705  RealScalar maxDiagEntry = m_workMatrix.cwiseAbs().diagonal().maxCoeff();
706 
707  bool finished = false;
708  while(!finished)
709  {
710  finished = true;
711 
712  // do a sweep: for all index pairs (p,q), perform SVD of the corresponding 2x2 sub-matrix
713 
714  for(Index p = 1; p < m_diagSize; ++p)
715  {
716  for(Index q = 0; q < p; ++q)
717  {
718  // if this 2x2 sub-matrix is not diagonal already...
719  // notice that this comparison will evaluate to false if any NaN is involved, ensuring that NaN's don't
720  // keep us iterating forever. Similarly, small denormal numbers are considered zero.
721  RealScalar threshold = numext::maxi<RealScalar>(considerAsZero, precision * maxDiagEntry);
722  if(abs(m_workMatrix.coeff(p,q))>threshold || abs(m_workMatrix.coeff(q,p)) > threshold)
723  {
724  finished = false;
725  // perform SVD decomposition of 2x2 sub-matrix corresponding to indices p,q to make it diagonal
726  // the complex to real operation returns true if the updated 2x2 block is not already diagonal
728  {
729  JacobiRotation<RealScalar> j_left, j_right;
730  internal::real_2x2_jacobi_svd(m_workMatrix, p, q, &j_left, &j_right);
731 
732  // accumulate resulting Jacobi rotations
733  m_workMatrix.applyOnTheLeft(p,q,j_left);
734  if(computeU()) m_matrixU.applyOnTheRight(p,q,j_left.transpose());
735 
736  m_workMatrix.applyOnTheRight(p,q,j_right);
737  if(computeV()) m_matrixV.applyOnTheRight(p,q,j_right);
738 
739  // keep track of the largest diagonal coefficient
740  maxDiagEntry = numext::maxi<RealScalar>(maxDiagEntry,numext::maxi<RealScalar>(abs(m_workMatrix.coeff(p,p)), abs(m_workMatrix.coeff(q,q))));
741  }
742  }
743  }
744  }
745  }
746 
747  /*** step 3. The work matrix is now diagonal, so ensure it's positive so its diagonal entries are the singular values ***/
748 
749  for(Index i = 0; i < m_diagSize; ++i)
750  {
751  // For a complex matrix, some diagonal coefficients might note have been
752  // treated by svd_precondition_2x2_block_to_be_real, and the imaginary part
753  // of some diagonal entry might not be null.
754  if(NumTraits<Scalar>::IsComplex && abs(numext::imag(m_workMatrix.coeff(i,i)))>considerAsZero)
755  {
756  RealScalar a = abs(m_workMatrix.coeff(i,i));
757  m_singularValues.coeffRef(i) = abs(a);
758  if(computeU()) m_matrixU.col(i) *= m_workMatrix.coeff(i,i)/a;
759  }
760  else
761  {
762  // m_workMatrix.coeff(i,i) is already real, no difficulty:
763  RealScalar a = numext::real(m_workMatrix.coeff(i,i));
764  m_singularValues.coeffRef(i) = abs(a);
765  if(computeU() && (a<RealScalar(0))) m_matrixU.col(i) = -m_matrixU.col(i);
766  }
767  }
768 
769  m_singularValues *= scale;
770 
771  /*** step 4. Sort singular values in descending order and compute the number of nonzero singular values ***/
772 
773  m_nonzeroSingularValues = m_diagSize;
774  for(Index i = 0; i < m_diagSize; i++)
775  {
776  Index pos;
777  RealScalar maxRemainingSingularValue = m_singularValues.tail(m_diagSize-i).maxCoeff(&pos);
778  if(maxRemainingSingularValue == RealScalar(0))
779  {
780  m_nonzeroSingularValues = i;
781  break;
782  }
783  if(pos)
784  {
785  pos += i;
786  std::swap(m_singularValues.coeffRef(i), m_singularValues.coeffRef(pos));
787  if(computeU()) m_matrixU.col(pos).swap(m_matrixU.col(i));
788  if(computeV()) m_matrixV.col(pos).swap(m_matrixV.col(i));
789  }
790  }
791 
792  m_isInitialized = true;
793  return *this;
794 }
795 
803 template<typename Derived>
805 MatrixBase<Derived>::jacobiSvd(unsigned int computationOptions) const
806 {
807  return JacobiSVD<PlainObject>(*this, computationOptions);
808 }
809 
810 } // end namespace Eigen
811 
812 #endif // EIGEN_JACOBISVD_H
EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const AbsReturnType abs() const
Definition: ArrayCwiseUnaryOps.h:52
EIGEN_DEVICE_FUNC const SqrtReturnType sqrt() const
Definition: ArrayCwiseUnaryOps.h:187
EIGEN_DEVICE_FUNC RealReturnType real() const
Definition: CommonCwiseUnaryOps.h:100
EIGEN_DEVICE_FUNC const ImagReturnType imag() const
Definition: CommonCwiseUnaryOps.h:109
#define EIGEN_SIZE_MIN_PREFER_DYNAMIC(a, b)
Definition: Macros.h:1294
#define eigen_assert(x)
Definition: Macros.h:1037
#define EIGEN_IMPLIES(a, b)
Definition: Macros.h:1315
#define EIGEN_SIZE_MIN_PREFER_FIXED(a, b)
Definition: Macros.h:1302
NumTraits< Scalar >::Real RealScalar
Definition: DenseBase.h:73
internal::traits< MatrixWrapper< ExpressionType > >::Scalar Scalar
Definition: DenseBase.h:66
Rotation given by a cosine-sine pair.
Definition: Jacobi.h:35
EIGEN_DEVICE_FUNC Scalar & s()
Definition: Jacobi.h:49
EIGEN_DEVICE_FUNC JacobiRotation transpose() const
Definition: Jacobi.h:63
EIGEN_DEVICE_FUNC Scalar & c()
Definition: Jacobi.h:47
EIGEN_DEVICE_FUNC JacobiRotation adjoint() const
Definition: Jacobi.h:67
Two-sided Jacobi SVD decomposition of a rectangular matrix.
Definition: JacobiSVD.h:490
Base::MatrixVType MatrixVType
Definition: JacobiSVD.h:508
MatrixType m_scaledMatrix
Definition: JacobiSVD.h:611
internal::plain_row_type< MatrixType >::type RowType
Definition: JacobiSVD.h:511
Matrix< Scalar, DiagSizeAtCompileTime, DiagSizeAtCompileTime, MatrixOptions, MaxDiagSizeAtCompileTime, MaxDiagSizeAtCompileTime > WorkMatrixType
Definition: JacobiSVD.h:515
_MatrixType MatrixType
Definition: JacobiSVD.h:494
NumTraits< typename MatrixType::Scalar >::Real RealScalar
Definition: JacobiSVD.h:496
void allocate(Index rows, Index cols, unsigned int computationOptions)
Definition: JacobiSVD.h:615
SVDBase< JacobiSVD > Base
Definition: JacobiSVD.h:491
internal::qr_preconditioner_impl< MatrixType, QRPreconditioner, internal::PreconditionIfMoreColsThanRows > m_qr_precond_morecols
Definition: JacobiSVD.h:609
WorkMatrixType m_workMatrix
Definition: JacobiSVD.h:602
EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR Index cols() const EIGEN_NOEXCEPT
Definition: EigenBase.h:63
JacobiSVD()
Default Constructor.
Definition: JacobiSVD.h:522
JacobiSVD(Index rows, Index cols, unsigned int computationOptions=0)
Default Constructor with memory preallocation.
Definition: JacobiSVD.h:532
JacobiSVD & compute(const MatrixType &matrix, unsigned int computationOptions)
Method performing the decomposition of given matrix using custom options.
Definition: JacobiSVD.h:666
Base::MatrixUType MatrixUType
Definition: JacobiSVD.h:507
internal::plain_col_type< MatrixType >::type ColType
Definition: JacobiSVD.h:512
@ MaxRowsAtCompileTime
Definition: JacobiSVD.h:501
@ MaxDiagSizeAtCompileTime
Definition: JacobiSVD.h:503
@ MaxColsAtCompileTime
Definition: JacobiSVD.h:502
@ ColsAtCompileTime
Definition: JacobiSVD.h:499
@ RowsAtCompileTime
Definition: JacobiSVD.h:498
@ DiagSizeAtCompileTime
Definition: JacobiSVD.h:500
@ MatrixOptions
Definition: JacobiSVD.h:504
Base::SingularValuesType SingularValuesType
Definition: JacobiSVD.h:509
JacobiSVD(const MatrixType &matrix, unsigned int computationOptions=0)
Constructor performing the decomposition of given matrix.
Definition: JacobiSVD.h:547
internal::qr_preconditioner_impl< MatrixType, QRPreconditioner, internal::PreconditionIfMoreRowsThanCols > m_qr_precond_morerows
Definition: JacobiSVD.h:610
JacobiSVD & compute(const MatrixType &matrix)
Method performing the decomposition of given matrix using current options.
Definition: JacobiSVD.h:570
EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR Index rows() const EIGEN_NOEXCEPT
Definition: EigenBase.h:60
MatrixType::Scalar Scalar
Definition: JacobiSVD.h:495
JacobiSVD< PlainObject > jacobiSvd(unsigned int computationOptions=0) const
Definition: JacobiSVD.h:805
EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Scalar & coeffRef(Index rowId, Index colId)
Definition: PlainObjectBase.h:175
EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE const Scalar & coeff(Index rowId, Index colId) const
Definition: PlainObjectBase.h:152
Base class of SVD algorithms.
Definition: SVDBase.h:64
Index cols() const
Definition: SVDBase.h:213
bool m_usePrescribedThreshold
Definition: SVDBase.h:276
bool m_computeFullV
Definition: SVDBase.h:278
Index rank() const
Definition: SVDBase.h:148
ComputationInfo m_info
Definition: SVDBase.h:275
bool m_computeThinU
Definition: SVDBase.h:277
bool computeV() const
Definition: SVDBase.h:210
bool m_isInitialized
Definition: SVDBase.h:276
Index m_cols
Definition: SVDBase.h:280
unsigned int m_computationOptions
Definition: SVDBase.h:279
MatrixVType m_matrixV
Definition: SVDBase.h:273
Index m_diagSize
Definition: SVDBase.h:280
bool computeU() const
Definition: SVDBase.h:208
internal::plain_diag_type< MatrixType, RealScalar >::type SingularValuesType
Definition: SVDBase.h:87
Index m_rows
Definition: SVDBase.h:280
Index m_nonzeroSingularValues
Definition: SVDBase.h:280
Index rows() const
Definition: SVDBase.h:212
Matrix< Scalar, RowsAtCompileTime, RowsAtCompileTime, MatrixOptions, MaxRowsAtCompileTime, MaxRowsAtCompileTime > MatrixUType
Definition: SVDBase.h:85
SingularValuesType m_singularValues
Definition: SVDBase.h:274
RealScalar m_prescribedThreshold
Definition: SVDBase.h:281
bool m_computeThinV
Definition: SVDBase.h:278
bool m_computeFullU
Definition: SVDBase.h:277
MatrixUType m_matrixU
Definition: SVDBase.h:272
Matrix< Scalar, ColsAtCompileTime, ColsAtCompileTime, MatrixOptions, MaxColsAtCompileTime, MaxColsAtCompileTime > MatrixVType
Definition: SVDBase.h:86
bool m_isAllocated
Definition: SVDBase.h:276
Definition: XprHelper.h:258
bool run(JacobiSVD< MatrixType, ColPivHouseholderQRPreconditioner > &svd, const MatrixType &matrix)
Definition: JacobiSVD.h:224
internal::make_proper_matrix_type< Scalar, ColsAtCompileTime, RowsAtCompileTime, Options, MaxColsAtCompileTime, MaxRowsAtCompileTime >::type TransposeTypeWithSameStorageOrder
Definition: JacobiSVD.h:210
ColPivHouseholderQR< TransposeTypeWithSameStorageOrder > QRType
Definition: JacobiSVD.h:245
void allocate(const JacobiSVD< MatrixType, ColPivHouseholderQRPreconditioner > &svd)
Definition: JacobiSVD.h:212
void allocate(const JacobiSVD< MatrixType, ColPivHouseholderQRPreconditioner > &svd)
Definition: JacobiSVD.h:159
bool run(JacobiSVD< MatrixType, ColPivHouseholderQRPreconditioner > &svd, const MatrixType &matrix)
Definition: JacobiSVD.h:170
bool run(JacobiSVD< MatrixType, FullPivHouseholderQRPreconditioner > &svd, const MatrixType &matrix)
Definition: JacobiSVD.h:133
void allocate(const JacobiSVD< MatrixType, FullPivHouseholderQRPreconditioner > &svd)
Definition: JacobiSVD.h:122
internal::make_proper_matrix_type< Scalar, ColsAtCompileTime, RowsAtCompileTime, Options, MaxColsAtCompileTime, MaxRowsAtCompileTime >::type TransposeTypeWithSameStorageOrder
Definition: JacobiSVD.h:120
FullPivHouseholderQR< TransposeTypeWithSameStorageOrder > QRType
Definition: JacobiSVD.h:147
void allocate(const JacobiSVD< MatrixType, FullPivHouseholderQRPreconditioner > &svd)
Definition: JacobiSVD.h:76
bool run(JacobiSVD< MatrixType, FullPivHouseholderQRPreconditioner > &svd, const MatrixType &matrix)
Definition: JacobiSVD.h:86
Matrix< Scalar, 1, RowsAtCompileTime, RowMajor, 1, MaxRowsAtCompileTime > WorkspaceType
Definition: JacobiSVD.h:74
bool run(JacobiSVD< MatrixType, HouseholderQRPreconditioner > &svd, const MatrixType &matrix)
Definition: JacobiSVD.h:268
void allocate(const JacobiSVD< MatrixType, HouseholderQRPreconditioner > &svd)
Definition: JacobiSVD.h:257
bool run(JacobiSVD< MatrixType, HouseholderQRPreconditioner > &svd, const MatrixType &matrix)
Definition: JacobiSVD.h:321
void allocate(const JacobiSVD< MatrixType, HouseholderQRPreconditioner > &svd)
Definition: JacobiSVD.h:309
internal::make_proper_matrix_type< Scalar, ColsAtCompileTime, RowsAtCompileTime, Options, MaxColsAtCompileTime, MaxRowsAtCompileTime >::type TransposeTypeWithSameStorageOrder
Definition: JacobiSVD.h:307
void allocate(const JacobiSVD< MatrixType, QRPreconditioner > &)
Definition: JacobiSVD.h:55
bool run(JacobiSVD< MatrixType, QRPreconditioner > &, const MatrixType &)
Definition: JacobiSVD.h:56
@ pos
Definition: Typedefs.h:19
@ NoQRPreconditioner
Definition: Constants.h:425
@ HouseholderQRPreconditioner
Definition: Constants.h:427
@ ColPivHouseholderQRPreconditioner
Definition: Constants.h:429
@ FullPivHouseholderQRPreconditioner
Definition: Constants.h:431
@ InvalidInput
Definition: Constants.h:449
@ Success
Definition: Constants.h:442
@ ComputeFullV
Definition: Constants.h:397
@ ComputeThinV
Definition: Constants.h:399
@ ComputeFullU
Definition: Constants.h:393
@ ComputeThinU
Definition: Constants.h:395
EIGEN_STRONG_INLINE EIGEN_DEVICE_FUNC bfloat16() min(const bfloat16 &a, const bfloat16 &b)
Definition: BFloat16.h:571
void real_2x2_jacobi_svd(const MatrixType &matrix, Index p, Index q, JacobiRotation< RealScalar > *j_left, JacobiRotation< RealScalar > *j_right)
Definition: RealSvd2x2.h:19
@ PreconditionIfMoreColsThanRows
Definition: JacobiSVD.h:30
@ PreconditionIfMoreRowsThanCols
Definition: JacobiSVD.h:30
EIGEN_DEVICE_FUNC EIGEN_ALWAYS_INLINE bool() isfinite(const Eigen::bfloat16 &h)
Definition: BFloat16.h:671
EIGEN_DEVICE_FUNC bool abs2(bool x)
Definition: MathFunctions.h:1292
Namespace containing all symbols from the Eigen library.
Definition: LDLT.h:16
EIGEN_DEFAULT_DENSE_INDEX_TYPE Index
The Index type as used for the API.
Definition: Meta.h:74
const int Dynamic
Definition: Constants.h:22
type
The type the bitset is encoded with.
Definition: bitset.hpp:44
Definition: document.h:416
NLOHMANN_BASIC_JSON_TPL_DECLARATION void swap(nlohmann::NLOHMANN_BASIC_JSON_TPL &j1, nlohmann::NLOHMANN_BASIC_JSON_TPL &j2) noexcept(//NOLINT(readability-inconsistent-declaration-parameter-name, cert-dcl58-cpp) is_nothrow_move_constructible< nlohmann::NLOHMANN_BASIC_JSON_TPL >::value &&//NOLINT(misc-redundant-expression, cppcoreguidelines-noexcept-swap, performance-noexcept-swap) is_nothrow_move_assignable< nlohmann::NLOHMANN_BASIC_JSON_TPL >::value)
exchanges the values of two JSON objects
Definition: json.hpp:25399
const GenericPointer< typename T::ValueType > T2 T::AllocatorType & a
Definition: pointer.h:1181
Eigen::Index Index
The interface type of indices.
Definition: EigenBase.h:39
@ IsComplex
Definition: NumTraits.h:157
Holds information about the various numeric (i.e. scalar) types allowed by Eigen.
Definition: NumTraits.h:233
Definition: JacobiSVD.h:49
JacobiSVD< MatrixType, QRPreconditioner > SVD
Definition: JacobiSVD.h:356
static bool run(typename SVD::WorkMatrixType &, SVD &, Index, Index, RealScalar &)
Definition: JacobiSVD.h:358
JacobiSVD< MatrixType, QRPreconditioner > SVD
Definition: JacobiSVD.h:364
static bool run(typename SVD::WorkMatrixType &work_matrix, SVD &svd, Index p, Index q, RealScalar &maxDiagEntry)
Definition: JacobiSVD.h:367
Definition: ForwardDeclarations.h:17
Definition: Meta.h:96