MatrixLinearSolve

MatrixLinearSolve [/D={subDiagonals,superDiagonals }/M=method /O/Z] [/L][/U] matrixA matrixB

The MatrixLinearSolve operation solves the linear system matrixA *X=matrixB where matrixA is an N-by-N matrix and matrixB is an N-by-NRHS matrix of the same data type.

Flags

/M=method

Determine the solution method which best suites input matrixA.

method =1:	Uses simple LU decomposition (default). See also LAPACK documentation for SGESV, CGESV, DGESV, and ZGESV.
	Creates the wave W_IPIV that contains the pivot indices that define the permutation matrix P. Row (i) if the matrix was interchanged with row ipiv(i).
method =2:	If matrixA is band diagonal, you also have to specify /D. See also LAPACK documentation for SGBSV, CGBSV, DGBSV, and ZGBSV.
	Creates the wave W_IPIV that contains the pivot indices that define the permutation matrix P. Row (i) if the matrix was interchanged with row ipiv(i). Also note that if you are using the /O flag, the overwritten waves may have a different dimensions.
method =4:	For tridiagonal matrix; still expecting full matrix in matrixA, but it will ignore the data in the elements outside the 3 diagonals. See also LAPACK documentation for SGTSV, CGTSV, DGTSV, and ZGTSV.
method =8:	Symmetric/hermitian. See also LAPACK documentation for SPOSV, CPOSV, DPOSV, and ZPOSV.
method =16:	Complex symmetric (complex only). See also LAPACK documentation for CSYSV and ZSYSV.

/D={sub,super}

Specifies a band diagonal matrix. The subdiagonal (sub) and superdiagonal (super) size must be positive integers.

Uses the lower triangle of matrixA. /L and /U are mutually exclusive flags.

Uses the upper triangle of matrixA. /U is the default.

Overwrites matrixA and matrixB with the results of the operation. This will save on the amount of memory needed.

No error reporting.

Details

If /O is not specified, the operation also creates the n-by-n wave M_A and the n-by-nrhs solution wave M_B.

The variable V_flag is created by the operation. If the operation completes successfully, V_flag is set to zero, otherwise it is set to the LAPACK error code.