PLASMA
2.8.0
PLASMA - Parallel Linear Algebra for Scalable Multi-core Architectures
|
void CORE_dlaed3_computevectors | ( | int | K, |
int | il_nondef, | ||
int | iu_nondef, | ||
double * | Q, | ||
int | LDQ, | ||
double * | W, | ||
double * | S, | ||
const int * | INDXC, | ||
int | start, | ||
int | end | ||
) |
CORE_dlaed3_computevectors - Compute the subset(start:end) of eigenvectors of the modified rank-1 system.
[in] | K | K specifies the number of non-deflated eigenvalues |
[in] | il_nondef | The first eigenvector index when computing only a subset of all eigenpairs |
[in] | iu_nondef | The last eigenvector index when computing only a subset of all eigenpairs |
[in,out] | Q | Q is an array of dimension (LDQ, end) Initially the first K columns of the full Q matrix are used as workspace. On output the Q(start:end) contain the updated eigenvectors. |
[in] | LDQ | LDQ specifies the leading dimension of Q |
[in] | W | W is the result of the previous computation after reduction. See core_dlaed3_computeW() and core_dlaed3_reduceW(). The first K elements of this array contain the components of the deflation-adjusted updating vector. Destroyed on output. |
[in] | S | S is local workspace of size K. |
[in] | INDXC | INDXC is the permutation used to arrange the columns of the deflated Q matrix into three groups (see core_dlaed2_computeK()). The rows of the eigenvectors found by core_dlaed4() must be likewise permuted before the matrix multiply of core_dlaed3_updatevectors() can take place. |
[in] | start | start specifies the first column index to be considered by this kernel |
[in] | end | end specifies the last column index to be considered by this kernel |