Let $A,B$ be two $m\times n$ real matrices. Then $$|AA'|\cdot |BB'|\geq |AB'|^2.$$

For square matrices, it is the equality. How to prove this inequality then?

  • 1
    $\begingroup$ Hint: Cauchy-Schwarz inequality; especially in the case where these matrices are vectors. $\endgroup$ – Squirtle Mar 22 '14 at 0:11
  • $\begingroup$ By $ A^{'} $, do you mean transpose? $\endgroup$ – user122283 Mar 22 '14 at 0:12
  • $\begingroup$ I think you need square roots on the left-hand-side. $\endgroup$ – Martin Argerami Mar 22 '14 at 1:27
  • $\begingroup$ @Squirtle I do not see what is the norm and inner product... $\endgroup$ – XLDD Mar 22 '14 at 11:11

Hint. If $m>n$, both sides are zero. If $m\le n$, by singular value decomposition, you may assume that $A=\pmatrix{D&0}$ and $B=\pmatrix{X&Y}$ for some diagonal matrix $D$, square matrix $X$ and $m\times(n-m)$ matrix $Y$. Using the identities $\det(PQ)=\det(P)\det(Q)$ and $\det(P^T)=\det(P)$ for square matrices, you may rewrite the alleged equality as $\det[D(XX^T+YY^T)D]\ge\det(DXX^TD)$. How do the two positive semidefinite matrices $D(XX^T+YY^T)D$ and $DXX^TD$ compare?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.