← All problems

Diagonal matrices maximize log-minor variance

7 years old · traced to

Let Mκ\mathcal M_{\kappa} be the set of positive-definite n×nn\times n matrices with condition number κ\kappa. For k<nk<n, κ\kappa, and M∈MκM\in\mathcal M_{\kappa}, define Yk(M)Y_k(M) as the log-minor random variable associated with MM. Diagonal matrices maximize log-minor variance. For all k<nk<n, κ\kappa, and M∈MκM\in\mathcal M_{\kappa}, there exists a diagonal matrix D∈MκD\in\mathcal M_{\kappa} such that

var⁡(Yk(M))≤var⁡(Yk(D)).\operatorname{var}(Y_k(M))\leq\operatorname{var}(Y_k(D)).

The conjecture is motivated by the sharpness of the variance bound for diagonal matrices and examples suggesting that diagonal matrices attain the largest possible log-minor variance among positive-definite matrices with fixed condition number. Its resolution would identify the extremizers for the general variance bound.

References
  1. Alice C. Schwarze, Philip S. Chodrow and Mason A. Porter, “Log-minor distributions and an application to estimating mean subsystem entropy”, arXiv:1901.09456 (2019).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.