1
$\begingroup$

A magic unitary is a unitary matrix $u=(p_{ij})_{ij}$ whose entries are all projections (in some Hilbert space) and in each row they sum to the identity and same holds for each column $(\sum_i p_{ij}=1=\sum_j p_{ij})$. My question is, if there is a sequence of "almost magic unitaries", by which I mean matrices $u_k=(q_{ij}^k)$ whose entries are all projections BUT they only sum to identity in each row and NOT in each column but close to identity ($\sum_j q_{ij}^k=1$ and for each j, $\sum_i ||q_{ij}^k-1||_2\rightarrow 0$) such that $u_k\rightarrow u$ in $||.||_2$ norm, Can one find actual magic unitaries that are close to this "almost magic unitaries" which approximate the original magic unitary?

It's similar to finding an orthogonal set of projections summing to identity from a set of almost orthogonal positive operators almost adding up to the identity.

$\endgroup$
  • $\begingroup$ Are these matrices square? finite? $\endgroup$ – Gerry Myerson Apr 30 at 4:22
  • $\begingroup$ Could you clarify what the norm $\|\cdot\|_2$ is? $\endgroup$ – Matthew Daws Apr 30 at 8:18
  • 1
    $\begingroup$ These matrices are all square matrices. The $||.||_2$ norm is the trace norm (Frobenius norm) $||x||_2=\sqrt {tr(xx^*)}$. Here we are assuming the projections live in a finite dimensional Hilbert space where the "tr"(trace functional) makes sense. $\endgroup$ – Miza Apr 30 at 19:40

Your Answer

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

Browse other questions tagged or ask your own question.