Dear Sumant, I suspect that (some of) your matrices are ill-conditioned, which would explain why sometimes dposv thinks your matrices are not positive definite. If this is the case, then you cannot expect different solvers to produce the same results. Perhaps you should think about how you are modelling your application and see if you can produce better conditioned matrices? Best wishes, Sven Hammarling. |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | [Lapack] xLARFP and scaling., Sven Hammarling |
|---|---|
| Next by Date: | [Lapack] xLARFP and scaling., Pat Quillen |
| Previous by Thread: | [Lapack] Clapack dgesv accuracy, Sumant Awate |
| Next by Thread: | [Lapack] xLARFP and scaling., Pat Quillen |
| Indexes: | [Date] [Thread] |