Skip to content
TheChatArchiveThe Chat Archive
o1-preview-2024-09-12· OpenAIAi2 WildChat-4.8M (AI2)

How would I prove that the condition number of a PSD matrix is equal to the square root…

  1. User

    How would I prove that the condition number of a PSD matrix is equal to the square root of the condition number of that PSD matrix * it's transpose

  2. Assistant

    To prove that the condition number of a matrix ( A ) is equal to the square root of the condition number of ( A A^\top ), we can use the properties of singular values and the definition of the condition number in terms of the 2-norm.

    Definitions:

    • The singular values of a matrix ( A ) are the square roots of the eigenvalues of ( A^\top A ) (or ( A A^\top )), and they are always non-negative real numbers. Denote them by ( \sigma_1 \geq \sigma_2 \geq \cdots \geq \sigma_p \geq 0 ), where ( p = \min(m, n) ) for an ( m \times n ) matrix ( A ).

    • The condition number ( \kappa(A) ) of a matrix ( A ) with respect to the 2-norm is defined as: [ \kappa(A) = \frac{|A|2 |A^{-1}|2}{|I|2} = \frac{\sigma{\max}(A)}{\sigma{\min}(A)} ] where ( \sigma{\max}(A) ) and ( \sigma_{\min}(A) ) are the largest and smallest singular values of ( A ), respectively.

    Proof:

    1. Eigenvalues of ( A A^\top ):

      The eigenvalues ( \lambda_i ) of ( A A^\top ) are the squares of the singular values of ( A ): [ \lambda_i = \sigma_i^2 ]

    2. Condition number of ( A A^\top ):

      The condition number of ( A A^\top ) is: [ \kappa(A A^\top) = \frac{\lambda_{\max}}{\lambda_{\min}} = \frac{\sigma_{\max}^2}{\sigma_{\min}^2} = \left( \frac{\sigma_{\max}}{\sigma_{\min}} \right)^2 = [\kappa(A)]^2 ]

    3. Relation between ( \kappa(A) ) and ( \kappa(A A^\top) ):

      Taking the square root of both sides: [ \kappa(A) = \sqrt{\kappa(A A^\top)} ]

    Conclusion:

    The condition number of a matrix ( A ) is equal to the square root of the condition number of ( A A^\top ), because the condition number squared of ( A ) gives the condition number of ( A A^\top ).

    Note: This result holds for any matrix ( A ), not necessarily symmetric or positive definite, as long as the singular values are defined (which they are for any real or complex matrix).

    Answer: By showing that κ(A Aᵗ) equals [κ(A)]², so κ(A) = √[κ(A Aᵗ)]; that is, the condition number of A equals the square root of that of A Aᵗ.