Currently matrixwrapper exposes a SymmetricMatrix class.
Would a Positive Definite Symmetric Matrix class be more useful ?
This would allow to calculate e.g. the inverse and the determinant more
efficiently (with cholesky decomposition).
Again, I can't imagine an application for BFL using negative semidefinite
symmetric matrices.
Regards,
Erwin
_______________________________________________
I hereby promise not to top-post on the
BFL mailing list
BFL [..] ...
http://lists.mech.kuleuven.be/mailman/listinfo/bfl
Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm
Matrix wrapper
On 5/8/07, Erwin Aertbelien <Erwin [dot] Aertbelien [..] ...> wrote:
> Currently matrixwrapper exposes a SymmetricMatrix class.
> Would a Positive Definite Symmetric Matrix class be more useful ?
Yes, the reason why I took SymmetricMatrix instead (at the time being,
that must have been somewhere in 2001 :) is the fact that most matrix
libraries at that time didn't provide a "pdsm" class, and BFL's first
concern was not about the most efficient implementation.
regards,
Klaas
_______________________________________________
I hereby promise not to top-post on the
BFL mailing list
BFL [..] ...
http://lists.mech.kuleuven.be/mailman/listinfo/bfl