?stedc. Computes the Z vector determining the rank-one modification of the diagonal matrix. Used when the original matrix is dense.
Zvector corresponding to the merge step in the
curlvl-th step of the merge process with
tlvlssteps for the
- INTEGER. The dimension of the symmetric tridiagonal matrix ().n≥0
- INTEGER. The total number of merging levels in the overall divide and conquer tree.
- INTEGER. The current level in the overall merge routine,0 ≤.curlvl≤tlvls
- INTEGER. The current problem in the current level in the overall merge routine (counting from upper left to lower right).
- INTEGER. Arrays, dimension (nlog2n) each.The arrayprmptr(*) contains a list of pointers which indicate where inperma level's permutation is stored.indicates the size of the permutation and also the size of the full, non-deflated problem.prmptr(i+1) -prmptr(i)The arrayperm(*) contains the permutations (from deflation and sorting) to be applied to each eigenblock.The arraygivptr(*) contains a list of pointers which indicate where ingivcola level's Givens rotations are stored.indicates the number of Givens rotations.givptr(i+1) -givptr(i)
- INTEGER. Array, dimension (2,nlog2n).Each pair of numbers indicates a pair of columns to take place in a Givens rotation.
- REALforslaedaDOUBLE PRECISIONfordlaeda.Array, dimension (2,nlog2n).Each number indicates theSvalue to be used in the corresponding Givens rotation.
- REALforslaedaDOUBLE PRECISIONfordlaeda.Array, dimension (n2).Contains the square eigenblocks from previous levels, the starting positions for blocks are given byqptr.
- INTEGER. Array, dimension (n+2). Contains a list of pointers which indicate where inqan eigenblock is stored.) indicates the size of the block.sqrt(qptr(i+1) -qptr(i)
- REALforslaedaDOUBLE PRECISIONfordlaeda.Workspace array, dimension (n).
- REALforslaedaDOUBLE PRECISIONfordlaeda.Array, dimension (