SVDErrorType#
The error type used to assess accuracy of computed solutions.
Synopsis#
typedef enum { SVD_ERROR_ABSOLUTE,
SVD_ERROR_RELATIVE,
SVD_ERROR_NORM } SVDErrorType;
Values#
SVD_ERROR_ABSOLUTE- compute error bound as \(\|r\|\)SVD_ERROR_RELATIVE- compute error bound as \(\|r\|/\sigma\)SVD_ERROR_NORM- compute error bound as \(\|r\|/\max\{\|A\|,\|B\|\}\)
Note#
The residual norm \(\|r\|\) is actually computed from two parts, such as
\(\sqrt{\eta_1^2+\eta_2^2}\) with \(\eta_1 = \|Av-\sigma u\|_2\) and
\(\eta_2 = \|A^*u-\sigma v\|_2\), see more details at SVDComputeError().
There is also a normalization factor related to the norm of the vectors,
which also varies with the problem type.
See Also#
SVD: Singular Value Decomposition, SVDComputeError(), SVDProblemType
Level#
intermediate
Location#
Index of all SVD routines Table of Contents for all manual pages Index of all manual pages