par

Type Members

1. class Fac_Cholesky extends Factorization with Error

   

   The Fac_Cholesky class provides methods to factor an 'n-by-n' symmetric, positive definite matrix 'a' into the product of two matrices:

   The Fac_Cholesky class provides methods to factor an 'n-by-n' symmetric, positive definite matrix 'a' into the product of two matrices:

   'l' - an 'n-by-n' left lower triangular matrix 'l.t' - an 'n-by-n' right upper triangular matrix - transpose of 'l'

   such that 'a = l * l.t'. This version uses parallel processing to speed up execution.

2. class Fac_QR extends Factorization with Error

   

   The Fac_QR class provides methods to factor an 'm-by-n' matrix 'a' into the product of two matrices:

   The Fac_QR class provides methods to factor an 'm-by-n' matrix 'a' into the product of two matrices:

   'q' - an 'm-by-n' orthogonal matrix and 'r' - an 'n-by-n' right upper triangular matrix

   such that 'a = q * r'. It uses Gram-Schmidt orthogonalization. Note, orthogonal means that 'q.t * q = I'. This version uses parallel processing to speed up execution.

   See also

   http://en.wikipedia.org/wiki/Gram–Schmidt_process (stabilized Gram–Schmidt orthonormalization)

   http://www.stat.wisc.edu/~larget/math496/qr.html

3. class MatrixD extends Matrix with Error with Serializable

   

   The MatrixD class stores and operates parallel on Numeric Matrices of type Double.

   The MatrixD class stores and operates parallel on Numeric Matrices of type Double. This class follows the MatrixN framework and is provided for efficiency. This class is only efficient when the dimension is large.

4. class ParSparseMatrixD extends Matrix with Error with Serializable

   

   The ParSparseMatrixD class stores and operates on Matrices of Doubles.

   The ParSparseMatrixD class stores and operates on Matrices of Doubles. Rather than storing the matrix as a 2 dimensional array, it is stored as an array of sorted-linked-maps, which record all the non-zero values for each particular row, along with their j-index as (j, v) pairs. Note: _npp versions of methods are not appropriate for sparse matrices (i.e., always use partial pivoting).