Given a symmetric matrix mP, what would be the fastest way to calculate the matrix quadratic form vX' * mP * vX?
Will it give extra performance if mP is positive definite?
Please address both the dense and the sparse case.
Given a symmetric matrix mP, what would be the fastest way to calculate the matrix quadratic form vX' * mP * vX?
Will it give extra performance if mP is positive definite?
Please address both the dense and the sparse case.