Public Types | Public Member Functions | Protected Types | Protected Attributes
SparseLLT Class Reference

LLT Cholesky decomposition of a sparse matrix and associated features. More...

List of all members.

Public Types

typedef SparseMatrix< Scalar > CholMatrixType
typedef MatrixType::Index Index
typedef _MatrixType MatrixType

Public Member Functions

Index cols () const
void compute (const MatrixType &matrix)
int flags () const
const CholMatrixTypematrixL (void) const
RealScalar precision () const
Index rows () const
void setFlags (int f)
void setPrecision (RealScalar v)
template<typename Rhs >
const internal::solve_retval
< SparseLLT< MatrixType >, Rhs > 
solve (const MatrixBase< Rhs > &b) const
template<typename Derived >
bool solveInPlace (MatrixBase< Derived > &b) const
 SparseLLT (int flags=0)
 SparseLLT (const MatrixType &matrix, int flags=0)
bool succeeded (void) const

Protected Types

enum  {
  SupernodalFactorIsDirty,
  MatrixLIsDirty
}
typedef NumTraits< typename
_MatrixType::Scalar >::Real 
RealScalar
typedef _MatrixType::Scalar Scalar

Protected Attributes

int m_flags
CholMatrixType m_matrix
RealScalar m_precision
int m_status
bool m_succeeded

Detailed Description

LLT Cholesky decomposition of a sparse matrix and associated features.

Parameters:
MatrixTypethe type of the matrix of which we are computing the LLT Cholesky decomposition
See also:
class LLT, class LDLT

Constructor & Destructor Documentation

SparseLLT ( int  flags = 0) [inline]

Creates a dummy LLT factorization object with flags flags.

SparseLLT ( const MatrixType &  matrix,
int  flags = 0 
) [inline]

Creates a LLT object and compute the respective factorization of matrix using flags flags.


Member Function Documentation

void compute ( const MatrixType &  matrix)

Computes/re-computes the LLT factorization

Computes / recomputes the LLT decomposition of matrix a using the default algorithm.

int flags ( ) const [inline]
Returns:
the current flags
const CholMatrixType& matrixL ( void  ) const [inline]
Returns:
the lower triangular matrix L
RealScalar precision ( ) const [inline]
Returns:
the current precision.
See also:
setPrecision()
void setFlags ( int  f) [inline]

Sets the flags. Possible values are:

  • CompleteFactorization
  • IncompleteFactorization
  • MemoryEfficient (hint to use the memory most efficient method offered by the backend)
  • SupernodalMultifrontal (implies a complete factorization if supported by the backend, overloads the MemoryEfficient flags)
  • SupernodalLeftLooking (implies a complete factorization if supported by the backend, overloads the MemoryEfficient flags)
See also:
flags()
void setPrecision ( RealScalar  v) [inline]

Sets the relative threshold value used to prune zero coefficients during the decomposition.

Setting a value greater than zero speeds up computation, and yields to an imcomplete factorization with fewer non zero coefficients. Such approximate factors are especially useful to initialize an iterative solver.

Warning:
if precision is greater that zero, the LLT factorization is not guaranteed to succeed even if the matrix is positive definite.

Note that the exact meaning of this parameter might depends on the actual backend. Moreover, not all backends support this feature.

See also:
precision()
bool solveInPlace ( MatrixBase< Derived > &  b) const

Computes b = L^-T L^-1 b

bool succeeded ( void  ) const [inline]
Returns:
true if the factorization succeeded

The documentation for this class was generated from the following file: