tomoto 0.2.3 → 0.3.0

data/vendor/eigen/Eigen/src/Core/SelfAdjointView.h

@@ -10,7 +10,7 @@
 #ifndef EIGEN_SELFADJOINTMATRIX_H
 #define EIGEN_SELFADJOINTMATRIX_H
 
-namespace Eigen { 
+namespace Eigen {
 
 /** \class SelfAdjointView
   * \ingroup Core_Module
@@ -58,14 +58,15 @@ template<typename _MatrixType, unsigned int UpLo> class SelfAdjointView
     typedef MatrixTypeNestedCleaned NestedExpression;
 
     /** \brief The type of coefficients in this matrix */
-    typedef typename internal::traits<SelfAdjointView>::Scalar Scalar; 
+    typedef typename internal::traits<SelfAdjointView>::Scalar Scalar;
     typedef typename MatrixType::StorageIndex StorageIndex;
     typedef typename internal::remove_all<typename MatrixType::ConjugateReturnType>::type MatrixConjugateReturnType;
+    typedef SelfAdjointView<typename internal::add_const<MatrixType>::type, UpLo> ConstSelfAdjointView;
 
     enum {
       Mode = internal::traits<SelfAdjointView>::Mode,
       Flags = internal::traits<SelfAdjointView>::Flags,
-      TransposeMode = ((Mode & Upper) ? Lower : 0) | ((Mode & Lower) ? Upper : 0)
+      TransposeMode = ((int(Mode) & int(Upper)) ? Lower : 0) | ((int(Mode) & int(Lower)) ? Upper : 0)
     };
     typedef typename MatrixType::PlainObject PlainObject;
 
@@ -75,14 +76,14 @@ template<typename _MatrixType, unsigned int UpLo> class SelfAdjointView
       EIGEN_STATIC_ASSERT(UpLo==Lower || UpLo==Upper,SELFADJOINTVIEW_ACCEPTS_UPPER_AND_LOWER_MODE_ONLY);
     }
 
-    EIGEN_DEVICE_FUNC
-    inline Index rows() const { return m_matrix.rows(); }
-    EIGEN_DEVICE_FUNC
-    inline Index cols() const { return m_matrix.cols(); }
-    EIGEN_DEVICE_FUNC
-    inline Index outerStride() const { return m_matrix.outerStride(); }
-    EIGEN_DEVICE_FUNC
-    inline Index innerStride() const { return m_matrix.innerStride(); }
+    EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR
+    inline Index rows() const EIGEN_NOEXCEPT { return m_matrix.rows(); }
+    EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR
+    inline Index cols() const EIGEN_NOEXCEPT { return m_matrix.cols(); }
+    EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR
+    inline Index outerStride() const EIGEN_NOEXCEPT { return m_matrix.outerStride(); }
+    EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR
+    inline Index innerStride() const EIGEN_NOEXCEPT { return m_matrix.innerStride(); }
 
     /** \sa MatrixBase::coeff()
       * \warning the coordinates must fit into the referenced triangular part
@@ -131,7 +132,7 @@ template<typename _MatrixType, unsigned int UpLo> class SelfAdjointView
     {
       return Product<OtherDerived,SelfAdjointView>(lhs.derived(),rhs);
     }
-    
+
     friend EIGEN_DEVICE_FUNC
     const SelfAdjointView<const EIGEN_SCALAR_BINARYOP_EXPR_RETURN_TYPE(Scalar,MatrixType,product),UpLo>
     operator*(const Scalar& s, const SelfAdjointView& mat)
@@ -197,6 +198,18 @@ template<typename _MatrixType, unsigned int UpLo> class SelfAdjointView
     inline const ConjugateReturnType conjugate() const
     { return ConjugateReturnType(m_matrix.conjugate()); }
 
+    /** \returns an expression of the complex conjugate of \c *this if Cond==true,
+     *           returns \c *this otherwise.
+     */
+    template<bool Cond>
+    EIGEN_DEVICE_FUNC
+    inline typename internal::conditional<Cond,ConjugateReturnType,ConstSelfAdjointView>::type
+    conjugateIf() const
+    {
+      typedef typename internal::conditional<Cond,ConjugateReturnType,ConstSelfAdjointView>::type ReturnType;
+      return ReturnType(m_matrix.template conjugateIf<Cond>());
+    }
+
     typedef SelfAdjointView<const typename MatrixType::AdjointReturnType,TransposeMode> AdjointReturnType;
     /** \sa MatrixBase::adjoint() const */
     EIGEN_DEVICE_FUNC
@@ -287,17 +300,17 @@ protected:
   using Base::m_src;
   using Base::m_functor;
 public:
-  
+
   typedef typename Base::DstEvaluatorType DstEvaluatorType;
   typedef typename Base::SrcEvaluatorType SrcEvaluatorType;
   typedef typename Base::Scalar Scalar;
   typedef typename Base::AssignmentTraits AssignmentTraits;
-  
-  
+
+
   EIGEN_DEVICE_FUNC triangular_dense_assignment_kernel(DstEvaluatorType &dst, const SrcEvaluatorType &src, const Functor &func, DstXprType& dstExpr)
     : Base(dst, src, func, dstExpr)
   {}
-  
+
   EIGEN_DEVICE_FUNC void assignCoeff(Index row, Index col)
   {
     eigen_internal_assert(row!=col);
@@ -305,12 +318,12 @@ public:
     m_functor.assignCoeff(m_dst.coeffRef(row,col), tmp);
     m_functor.assignCoeff(m_dst.coeffRef(col,row), numext::conj(tmp));
   }
-  
+
   EIGEN_DEVICE_FUNC void assignDiagonalCoeff(Index id)
   {
     Base::assignCoeff(id,id);
   }
-  
+
   EIGEN_DEVICE_FUNC void assignOppositeCoeff(Index, Index)
   { eigen_internal_assert(false && "should never be called"); }
 };
@@ -324,7 +337,7 @@ public:
 /** This is the const version of MatrixBase::selfadjointView() */
 template<typename Derived>
 template<unsigned int UpLo>
-typename MatrixBase<Derived>::template ConstSelfAdjointViewReturnType<UpLo>::Type
+EIGEN_DEVICE_FUNC typename MatrixBase<Derived>::template ConstSelfAdjointViewReturnType<UpLo>::Type
 MatrixBase<Derived>::selfadjointView() const
 {
   return typename ConstSelfAdjointViewReturnType<UpLo>::Type(derived());
@@ -341,7 +354,7 @@ MatrixBase<Derived>::selfadjointView() const
   */
 template<typename Derived>
 template<unsigned int UpLo>
-typename MatrixBase<Derived>::template SelfAdjointViewReturnType<UpLo>::Type
+EIGEN_DEVICE_FUNC typename MatrixBase<Derived>::template SelfAdjointViewReturnType<UpLo>::Type
 MatrixBase<Derived>::selfadjointView()
 {
   return typename SelfAdjointViewReturnType<UpLo>::Type(derived());

data/vendor/eigen/Eigen/src/Core/Solve.h

@@ -13,13 +13,13 @@
 namespace Eigen {
 
 template<typename Decomposition, typename RhsType, typename StorageKind> class SolveImpl;
-  
+
 /** \class Solve
   * \ingroup Core_Module
   *
   * \brief Pseudo expression representing a solving operation
   *
-  * \tparam Decomposition the type of the matrix or decomposion object
+  * \tparam Decomposition the type of the matrix or decomposition object
   * \tparam Rhstype the type of the right-hand side
   *
   * This class represents an expression of A.solve(B)
@@ -64,13 +64,13 @@ class Solve : public SolveImpl<Decomposition,RhsType,typename internal::traits<R
 public:
   typedef typename internal::traits<Solve>::PlainObject PlainObject;
   typedef typename internal::traits<Solve>::StorageIndex StorageIndex;
-  
+
   Solve(const Decomposition &dec, const RhsType &rhs)
     : m_dec(dec), m_rhs(rhs)
   {}
-  
-  EIGEN_DEVICE_FUNC Index rows() const { return m_dec.cols(); }
-  EIGEN_DEVICE_FUNC Index cols() const { return m_rhs.cols(); }
+
+  EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR Index rows() const EIGEN_NOEXCEPT { return m_dec.cols(); }
+  EIGEN_DEVICE_FUNC EIGEN_CONSTEXPR Index cols() const EIGEN_NOEXCEPT { return m_rhs.cols(); }
 
   EIGEN_DEVICE_FUNC const Decomposition& dec() const { return m_dec; }
   EIGEN_DEVICE_FUNC const RhsType&       rhs() const { return m_rhs; }
@@ -87,14 +87,14 @@ class SolveImpl<Decomposition,RhsType,Dense>
   : public MatrixBase<Solve<Decomposition,RhsType> >
 {
   typedef Solve<Decomposition,RhsType> Derived;
-  
+
 public:
-  
+
   typedef MatrixBase<Solve<Decomposition,RhsType> > Base;
   EIGEN_DENSE_PUBLIC_INTERFACE(Derived)
 
 private:
-  
+
   Scalar coeff(Index row, Index col) const;
   Scalar coeff(Index i) const;
 };
@@ -119,15 +119,15 @@ struct evaluator<Solve<Decomposition,RhsType> >
   typedef evaluator<PlainObject> Base;
 
   enum { Flags = Base::Flags | EvalBeforeNestingBit };
-  
+
   EIGEN_DEVICE_FUNC explicit evaluator(const SolveType& solve)
     : m_result(solve.rows(), solve.cols())
   {
     ::new (static_cast<Base*>(this)) Base(m_result);
     solve.dec()._solve_impl(solve.rhs(), m_result);
   }
-  
-protected:  
+
+protected:
   PlainObject m_result;
 };
 
@@ -176,12 +176,12 @@ struct Assignment<DstXprType, Solve<CwiseUnaryOp<internal::scalar_conjugate_op<t
     Index dstCols = src.cols();
     if((dst.rows()!=dstRows) || (dst.cols()!=dstCols))
       dst.resize(dstRows, dstCols);
-    
+
     src.dec().nestedExpression().nestedExpression().template _solve_impl_transposed<true>(src.rhs(), dst);
   }
 };
 
-} // end namepsace internal
+} // end namespace internal
 
 } // end namespace Eigen
 

data/vendor/eigen/Eigen/src/Core/SolveTriangular.h

@@ -10,7 +10,7 @@
 #ifndef EIGEN_SOLVETRIANGULAR_H
 #define EIGEN_SOLVETRIANGULAR_H
 
-namespace Eigen { 
+namespace Eigen {
 
 namespace internal {
 
@@ -19,7 +19,7 @@ namespace internal {
 template<typename LhsScalar, typename RhsScalar, typename Index, int Side, int Mode, bool Conjugate, int StorageOrder>
 struct triangular_solve_vector;
 
-template <typename Scalar, typename Index, int Side, int Mode, bool Conjugate, int TriStorageOrder, int OtherStorageOrder>
+template <typename Scalar, typename Index, int Side, int Mode, bool Conjugate, int TriStorageOrder, int OtherStorageOrder, int OtherInnerStride>
 struct triangular_solve_matrix;
 
 // small helper struct extracting some traits on the underlying solver operation
@@ -54,7 +54,7 @@ struct triangular_solver_selector<Lhs,Rhs,Side,Mode,NoUnrolling,1>
   typedef blas_traits<Lhs> LhsProductTraits;
   typedef typename LhsProductTraits::ExtractType ActualLhsType;
   typedef Map<Matrix<RhsScalar,Dynamic,1>, Aligned> MappedRhs;
-  static void run(const Lhs& lhs, Rhs& rhs)
+  static EIGEN_DEVICE_FUNC void run(const Lhs& lhs, Rhs& rhs)
   {
     ActualLhsType actualLhs = LhsProductTraits::extract(lhs);
 
@@ -64,7 +64,7 @@ struct triangular_solver_selector<Lhs,Rhs,Side,Mode,NoUnrolling,1>
 
     ei_declare_aligned_stack_constructed_variable(RhsScalar,actualRhs,rhs.size(),
                                                   (useRhsDirectly ? rhs.data() : 0));
-                                                  
+
     if(!useRhsDirectly)
       MappedRhs(actualRhs,rhs.size()) = rhs;
 
@@ -85,7 +85,7 @@ struct triangular_solver_selector<Lhs,Rhs,Side,Mode,NoUnrolling,Dynamic>
   typedef blas_traits<Lhs> LhsProductTraits;
   typedef typename LhsProductTraits::DirectLinearAccessType ActualLhsType;
 
-  static void run(const Lhs& lhs, Rhs& rhs)
+  static EIGEN_DEVICE_FUNC void run(const Lhs& lhs, Rhs& rhs)
   {
     typename internal::add_const_on_value_type<ActualLhsType>::type actualLhs = LhsProductTraits::extract(lhs);
 
@@ -98,8 +98,8 @@ struct triangular_solver_selector<Lhs,Rhs,Side,Mode,NoUnrolling,Dynamic>
     BlockingType blocking(rhs.rows(), rhs.cols(), size, 1, false);
 
     triangular_solve_matrix<Scalar,Index,Side,Mode,LhsProductTraits::NeedToConjugate,(int(Lhs::Flags) & RowMajorBit) ? RowMajor : ColMajor,
-                               (Rhs::Flags&RowMajorBit) ? RowMajor : ColMajor>
-      ::run(size, othersize, &actualLhs.coeffRef(0,0), actualLhs.outerStride(), &rhs.coeffRef(0,0), rhs.outerStride(), blocking);
+                               (Rhs::Flags&RowMajorBit) ? RowMajor : ColMajor, Rhs::InnerStrideAtCompileTime>
+      ::run(size, othersize, &actualLhs.coeffRef(0,0), actualLhs.outerStride(), &rhs.coeffRef(0,0), rhs.innerStride(), rhs.outerStride(), blocking);
   }
 };
 
@@ -118,7 +118,7 @@ struct triangular_solver_unroller<Lhs,Rhs,Mode,LoopIndex,Size,false> {
     DiagIndex  = IsLower ? LoopIndex : Size - LoopIndex - 1,
     StartIndex = IsLower ? 0         : DiagIndex+1
   };
-  static void run(const Lhs& lhs, Rhs& rhs)
+  static EIGEN_DEVICE_FUNC void run(const Lhs& lhs, Rhs& rhs)
   {
     if (LoopIndex>0)
       rhs.coeffRef(DiagIndex) -= lhs.row(DiagIndex).template segment<LoopIndex>(StartIndex).transpose()
@@ -133,22 +133,22 @@ struct triangular_solver_unroller<Lhs,Rhs,Mode,LoopIndex,Size,false> {
 
 template<typename Lhs, typename Rhs, int Mode, int LoopIndex, int Size>
 struct triangular_solver_unroller<Lhs,Rhs,Mode,LoopIndex,Size,true> {
-  static void run(const Lhs&, Rhs&) {}
+  static EIGEN_DEVICE_FUNC void run(const Lhs&, Rhs&) {}
 };
 
 template<typename Lhs, typename Rhs, int Mode>
 struct triangular_solver_selector<Lhs,Rhs,OnTheLeft,Mode,CompleteUnrolling,1> {
-  static void run(const Lhs& lhs, Rhs& rhs)
+  static EIGEN_DEVICE_FUNC void run(const Lhs& lhs, Rhs& rhs)
   { triangular_solver_unroller<Lhs,Rhs,Mode,0,Rhs::SizeAtCompileTime>::run(lhs,rhs); }
 };
 
 template<typename Lhs, typename Rhs, int Mode>
 struct triangular_solver_selector<Lhs,Rhs,OnTheRight,Mode,CompleteUnrolling,1> {
-  static void run(const Lhs& lhs, Rhs& rhs)
+  static EIGEN_DEVICE_FUNC void run(const Lhs& lhs, Rhs& rhs)
   {
     Transpose<const Lhs> trLhs(lhs);
     Transpose<Rhs> trRhs(rhs);
-    
+
     triangular_solver_unroller<Transpose<const Lhs>,Transpose<Rhs>,
                               ((Mode&Upper)==Upper ? Lower : Upper) | (Mode&UnitDiag),
                               0,Rhs::SizeAtCompileTime>::run(trLhs,trRhs);
@@ -164,11 +164,11 @@ struct triangular_solver_selector<Lhs,Rhs,OnTheRight,Mode,CompleteUnrolling,1> {
 #ifndef EIGEN_PARSED_BY_DOXYGEN
 template<typename MatrixType, unsigned int Mode>
 template<int Side, typename OtherDerived>
-void TriangularViewImpl<MatrixType,Mode,Dense>::solveInPlace(const MatrixBase<OtherDerived>& _other) const
+EIGEN_DEVICE_FUNC void TriangularViewImpl<MatrixType,Mode,Dense>::solveInPlace(const MatrixBase<OtherDerived>& _other) const
 {
   OtherDerived& other = _other.const_cast_derived();
   eigen_assert( derived().cols() == derived().rows() && ((Side==OnTheLeft && derived().cols() == other.rows()) || (Side==OnTheRight && derived().cols() == other.cols())) );
-  eigen_assert((!(Mode & ZeroDiag)) && bool(Mode & (Upper|Lower)));
+  eigen_assert((!(int(Mode) & int(ZeroDiag))) && bool(int(Mode) & (int(Upper) | int(Lower))));
   // If solving for a 0x0 matrix, nothing to do, simply return.
   if (derived().cols() == 0)
     return;
@@ -213,8 +213,8 @@ template<int Side, typename TriangularType, typename Rhs> struct triangular_solv
     : m_triangularMatrix(tri), m_rhs(rhs)
   {}
 
-  inline Index rows() const { return m_rhs.rows(); }
-  inline Index cols() const { return m_rhs.cols(); }
+  inline EIGEN_CONSTEXPR Index rows() const EIGEN_NOEXCEPT { return m_rhs.rows(); }
+  inline EIGEN_CONSTEXPR Index cols() const EIGEN_NOEXCEPT { return m_rhs.cols(); }
 
   template<typename Dest> inline void evalTo(Dest& dst) const
   {

data/vendor/eigen/Eigen/src/Core/SolverBase.h

@@ -14,8 +14,35 @@ namespace Eigen {
 
 namespace internal {
 
+template<typename Derived>
+struct solve_assertion {
+    template<bool Transpose_, typename Rhs>
+    static void run(const Derived& solver, const Rhs& b) { solver.template _check_solve_assertion<Transpose_>(b); }
+};
+
+template<typename Derived>
+struct solve_assertion<Transpose<Derived> >
+{
+    typedef Transpose<Derived> type;
+
+    template<bool Transpose_, typename Rhs>
+    static void run(const type& transpose, const Rhs& b)
+    {
+        internal::solve_assertion<typename internal::remove_all<Derived>::type>::template run<true>(transpose.nestedExpression(), b);
+    }
+};
 
+template<typename Scalar, typename Derived>
+struct solve_assertion<CwiseUnaryOp<Eigen::internal::scalar_conjugate_op<Scalar>, const Transpose<Derived> > >
+{
+    typedef CwiseUnaryOp<Eigen::internal::scalar_conjugate_op<Scalar>, const Transpose<Derived> > type;
 
+    template<bool Transpose_, typename Rhs>
+    static void run(const type& adjoint, const Rhs& b)
+    {
+        internal::solve_assertion<typename internal::remove_all<Transpose<Derived> >::type>::template run<true>(adjoint.nestedExpression(), b);
+    }
+};
 } // end namespace internal
 
 /** \class SolverBase
@@ -35,7 +62,7 @@ namespace internal {
   *
   * \warning Currently, any other usage of transpose() and adjoint() are not supported and will produce compilation errors.
   *
-  * \sa class PartialPivLU, class FullPivLU
+  * \sa class PartialPivLU, class FullPivLU, class HouseholderQR, class ColPivHouseholderQR, class FullPivHouseholderQR, class CompleteOrthogonalDecomposition, class LLT, class LDLT, class SVDBase
   */
 template<typename Derived>
 class SolverBase : public EigenBase<Derived>
@@ -46,6 +73,9 @@ class SolverBase : public EigenBase<Derived>
     typedef typename internal::traits<Derived>::Scalar Scalar;
     typedef Scalar CoeffReturnType;
 
+    template<typename Derived_>
+    friend struct internal::solve_assertion;
+
     enum {
       RowsAtCompileTime = internal::traits<Derived>::RowsAtCompileTime,
       ColsAtCompileTime = internal::traits<Derived>::ColsAtCompileTime,
@@ -56,7 +86,8 @@ class SolverBase : public EigenBase<Derived>
       MaxSizeAtCompileTime = (internal::size_at_compile_time<internal::traits<Derived>::MaxRowsAtCompileTime,
                                                              internal::traits<Derived>::MaxColsAtCompileTime>::ret),
       IsVectorAtCompileTime = internal::traits<Derived>::MaxRowsAtCompileTime == 1
-                           || internal::traits<Derived>::MaxColsAtCompileTime == 1
+                           || internal::traits<Derived>::MaxColsAtCompileTime == 1,
+      NumDimensions = int(MaxSizeAtCompileTime) == 1 ? 0 : bool(IsVectorAtCompileTime) ? 1 : 2
     };
 
     /** Default constructor */
@@ -74,7 +105,7 @@ class SolverBase : public EigenBase<Derived>
     inline const Solve<Derived, Rhs>
     solve(const MatrixBase<Rhs>& b) const
     {
-      eigen_assert(derived().rows()==b.rows() && "solve(): invalid number of rows of the right hand side matrix b");
+      internal::solve_assertion<typename internal::remove_all<Derived>::type>::template run<false>(derived(), b);
       return Solve<Derived, Rhs>(derived(), b.derived());
     }
 
@@ -112,6 +143,13 @@ class SolverBase : public EigenBase<Derived>
     }
 
   protected:
+
+    template<bool Transpose_, typename Rhs>
+    void _check_solve_assertion(const Rhs& b) const {
+        EIGEN_ONLY_USED_FOR_DEBUG(b);
+        eigen_assert(derived().m_isInitialized && "Solver is not initialized.");
+        eigen_assert((Transpose_?derived().cols():derived().rows())==b.rows() && "SolverBase::solve(): invalid number of rows of the right hand side matrix b");
+    }
 };
 
 namespace internal {

data/vendor/eigen/Eigen/src/Core/StableNorm.h

@@ -50,6 +50,71 @@ inline void stable_norm_kernel(const ExpressionType& bl, Scalar& ssq, Scalar& sc
     ssq += (bl*invScale).squaredNorm();
 }
 
+template<typename VectorType, typename RealScalar>
+void stable_norm_impl_inner_step(const VectorType &vec, RealScalar& ssq, RealScalar& scale, RealScalar& invScale)
+{
+  typedef typename VectorType::Scalar Scalar;
+  const Index blockSize = 4096;
+  
+  typedef typename internal::nested_eval<VectorType,2>::type VectorTypeCopy;
+  typedef typename internal::remove_all<VectorTypeCopy>::type VectorTypeCopyClean;
+  const VectorTypeCopy copy(vec);
+  
+  enum {
+    CanAlign = (   (int(VectorTypeCopyClean::Flags)&DirectAccessBit)
+                || (int(internal::evaluator<VectorTypeCopyClean>::Alignment)>0) // FIXME Alignment)>0 might not be enough
+               ) && (blockSize*sizeof(Scalar)*2<EIGEN_STACK_ALLOCATION_LIMIT)
+                 && (EIGEN_MAX_STATIC_ALIGN_BYTES>0) // if we cannot allocate on the stack, then let's not bother about this optimization
+  };
+  typedef typename internal::conditional<CanAlign, Ref<const Matrix<Scalar,Dynamic,1,0,blockSize,1>, internal::evaluator<VectorTypeCopyClean>::Alignment>,
+                                                   typename VectorTypeCopyClean::ConstSegmentReturnType>::type SegmentWrapper;
+  Index n = vec.size();
+  
+  Index bi = internal::first_default_aligned(copy);
+  if (bi>0)
+    internal::stable_norm_kernel(copy.head(bi), ssq, scale, invScale);
+  for (; bi<n; bi+=blockSize)
+    internal::stable_norm_kernel(SegmentWrapper(copy.segment(bi,numext::mini(blockSize, n - bi))), ssq, scale, invScale);
+}
+
+template<typename VectorType>
+typename VectorType::RealScalar
+stable_norm_impl(const VectorType &vec, typename enable_if<VectorType::IsVectorAtCompileTime>::type* = 0 )
+{
+  using std::sqrt;
+  using std::abs;
+
+  Index n = vec.size();
+
+  if(n==1)
+    return abs(vec.coeff(0));
+
+  typedef typename VectorType::RealScalar RealScalar;
+  RealScalar scale(0);
+  RealScalar invScale(1);
+  RealScalar ssq(0); // sum of squares
+
+  stable_norm_impl_inner_step(vec, ssq, scale, invScale);
+  
+  return scale * sqrt(ssq);
+}
+
+template<typename MatrixType>
+typename MatrixType::RealScalar
+stable_norm_impl(const MatrixType &mat, typename enable_if<!MatrixType::IsVectorAtCompileTime>::type* = 0 )
+{
+  using std::sqrt;
+
+  typedef typename MatrixType::RealScalar RealScalar;
+  RealScalar scale(0);
+  RealScalar invScale(1);
+  RealScalar ssq(0); // sum of squares
+
+  for(Index j=0; j<mat.outerSize(); ++j)
+    stable_norm_impl_inner_step(mat.innerVector(j), ssq, scale, invScale);
+  return scale * sqrt(ssq);
+}
+
 template<typename Derived>
 inline typename NumTraits<typename traits<Derived>::Scalar>::Real
 blueNorm_impl(const EigenBase<Derived>& _vec)
@@ -58,52 +123,43 @@ blueNorm_impl(const EigenBase<Derived>& _vec)
   using std::pow;
   using std::sqrt;
   using std::abs;
+
+  // This program calculates the machine-dependent constants
+  // bl, b2, slm, s2m, relerr overfl
+  // from the "basic" machine-dependent numbers
+  // nbig, ibeta, it, iemin, iemax, rbig.
+  // The following define the basic machine-dependent constants.
+  // For portability, the PORT subprograms "ilmaeh" and "rlmach"
+  // are used. For any specific computer, each of the assignment
+  // statements can be replaced
+  static const int ibeta = std::numeric_limits<RealScalar>::radix;  // base for floating-point numbers
+  static const int it    = NumTraits<RealScalar>::digits();  // number of base-beta digits in mantissa
+  static const int iemin = NumTraits<RealScalar>::min_exponent();  // minimum exponent
+  static const int iemax = NumTraits<RealScalar>::max_exponent();  // maximum exponent
+  static const RealScalar rbig   = NumTraits<RealScalar>::highest();  // largest floating-point number
+  static const RealScalar b1     = RealScalar(pow(RealScalar(ibeta),RealScalar(-((1-iemin)/2))));  // lower boundary of midrange
+  static const RealScalar b2     = RealScalar(pow(RealScalar(ibeta),RealScalar((iemax + 1 - it)/2)));  // upper boundary of midrange
+  static const RealScalar s1m    = RealScalar(pow(RealScalar(ibeta),RealScalar((2-iemin)/2)));  // scaling factor for lower range
+  static const RealScalar s2m    = RealScalar(pow(RealScalar(ibeta),RealScalar(- ((iemax+it)/2))));  // scaling factor for upper range
+  static const RealScalar eps    = RealScalar(pow(double(ibeta), 1-it));
+  static const RealScalar relerr = sqrt(eps);  // tolerance for neglecting asml
+
   const Derived& vec(_vec.derived());
-  static bool initialized = false;
-  static RealScalar b1, b2, s1m, s2m, rbig, relerr;
-  if(!initialized)
-  {
-    int ibeta, it, iemin, iemax, iexp;
-    RealScalar eps;
-    // This program calculates the machine-dependent constants
-    // bl, b2, slm, s2m, relerr overfl
-    // from the "basic" machine-dependent numbers
-    // nbig, ibeta, it, iemin, iemax, rbig.
-    // The following define the basic machine-dependent constants.
-    // For portability, the PORT subprograms "ilmaeh" and "rlmach"
-    // are used. For any specific computer, each of the assignment
-    // statements can be replaced
-    ibeta = std::numeric_limits<RealScalar>::radix;                 // base for floating-point numbers
-    it    = std::numeric_limits<RealScalar>::digits;                // number of base-beta digits in mantissa
-    iemin = std::numeric_limits<RealScalar>::min_exponent;          // minimum exponent
-    iemax = std::numeric_limits<RealScalar>::max_exponent;          // maximum exponent
-    rbig  = (std::numeric_limits<RealScalar>::max)();               // largest floating-point number
-
-    iexp  = -((1-iemin)/2);
-    b1    = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));    // lower boundary of midrange
-    iexp  = (iemax + 1 - it)/2;
-    b2    = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));    // upper boundary of midrange
-
-    iexp  = (2-iemin)/2;
-    s1m   = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));    // scaling factor for lower range
-    iexp  = - ((iemax+it)/2);
-    s2m   = RealScalar(pow(RealScalar(ibeta),RealScalar(iexp)));    // scaling factor for upper range
-
-    eps     = RealScalar(pow(double(ibeta), 1-it));
-    relerr  = sqrt(eps);                                            // tolerance for neglecting asml
-    initialized = true;
-  }
   Index n = vec.size();
   RealScalar ab2 = b2 / RealScalar(n);
   RealScalar asml = RealScalar(0);
   RealScalar amed = RealScalar(0);
   RealScalar abig = RealScalar(0);
-  for(typename Derived::InnerIterator it(vec, 0); it; ++it)
+
+  for(Index j=0; j<vec.outerSize(); ++j)
   {
-    RealScalar ax = abs(it.value());
-    if(ax > ab2)     abig += numext::abs2(ax*s2m);
-    else if(ax < b1) asml += numext::abs2(ax*s1m);
-    else             amed += numext::abs2(ax);
+    for(typename Derived::InnerIterator iter(vec, j); iter; ++iter)
+    {
+      RealScalar ax = abs(iter.value());
+      if(ax > ab2)     abig += numext::abs2(ax*s2m);
+      else if(ax < b1) asml += numext::abs2(ax*s1m);
+      else             amed += numext::abs2(ax);
+    }
   }
   if(amed!=amed)
     return amed;  // we got a NaN
@@ -156,36 +212,7 @@ template<typename Derived>
 inline typename NumTraits<typename internal::traits<Derived>::Scalar>::Real
 MatrixBase<Derived>::stableNorm() const
 {
-  using std::sqrt;
-  using std::abs;
-  const Index blockSize = 4096;
-  RealScalar scale(0);
-  RealScalar invScale(1);
-  RealScalar ssq(0); // sum of square
-  
-  typedef typename internal::nested_eval<Derived,2>::type DerivedCopy;
-  typedef typename internal::remove_all<DerivedCopy>::type DerivedCopyClean;
-  const DerivedCopy copy(derived());
-  
-  enum {
-    CanAlign = (   (int(DerivedCopyClean::Flags)&DirectAccessBit)
-                || (int(internal::evaluator<DerivedCopyClean>::Alignment)>0) // FIXME Alignment)>0 might not be enough
-               ) && (blockSize*sizeof(Scalar)*2<EIGEN_STACK_ALLOCATION_LIMIT)
-                 && (EIGEN_MAX_STATIC_ALIGN_BYTES>0) // if we cannot allocate on the stack, then let's not bother about this optimization
-  };
-  typedef typename internal::conditional<CanAlign, Ref<const Matrix<Scalar,Dynamic,1,0,blockSize,1>, internal::evaluator<DerivedCopyClean>::Alignment>,
-                                                   typename DerivedCopyClean::ConstSegmentReturnType>::type SegmentWrapper;
-  Index n = size();
-  
-  if(n==1)
-    return abs(this->coeff(0));
-  
-  Index bi = internal::first_default_aligned(copy);
-  if (bi>0)
-    internal::stable_norm_kernel(copy.head(bi), ssq, scale, invScale);
-  for (; bi<n; bi+=blockSize)
-    internal::stable_norm_kernel(SegmentWrapper(copy.segment(bi,numext::mini(blockSize, n - bi))), ssq, scale, invScale);
-  return scale * sqrt(ssq);
+  return internal::stable_norm_impl(derived());
 }
 
 /** \returns the \em l2 norm of \c *this using the Blue's algorithm.
@@ -213,7 +240,10 @@ template<typename Derived>
 inline typename NumTraits<typename internal::traits<Derived>::Scalar>::Real
 MatrixBase<Derived>::hypotNorm() const
 {
-  return this->cwiseAbs().redux(internal::scalar_hypot_op<RealScalar>());
+  if(size()==1)
+    return numext::abs(coeff(0,0));
+  else
+    return this->cwiseAbs().redux(internal::scalar_hypot_op<RealScalar>());
 }
 
 } // end namespace Eigen