5
$\begingroup$

Is there a measure for matrix that is analogous to rank of the matrix, but it is continuous on matrix elements? Say, we could say the information in identity matrix $I_n$ is $n$, and when the off-diagonal elements change from 0 to 1, the information contained in the matrix reduces continuously.

Example: considering a matrix $A(a)$ \begin{pmatrix} 1&0&0&\\ 0&1&a\\ 0&a&1 \end{pmatrix}. How do I define an information measure $I(A(a))$ that is continuous to $a$ for matrix $A(a)$, so that $I(A(0)) = 3$ and $I(A(1)) = 2$? Rank is not continuous, Shannon information entropy on eigenvalues does not give desired values.

Improve this question
$\endgroup$
3
  • $\begingroup$ en.wikipedia.org/wiki/Von_Neumann_entropy $\endgroup$
    – Steve Huntsman
    Commented Jun 16, 2016 at 19:25
  • $\begingroup$ $S(\rho) = -tr(\rho\log(\rho))$ equals zero for any identity matrix while the dimension information lost; and $S(A(1))$ from my example does not reduce to S($I_2$). $\endgroup$
    – ahala
    Commented Jun 16, 2016 at 19:58
  • 1
    $\begingroup$ If I have not made a silly mistake, rank($A$) =rank($A^*A$) by some version of the polar decomposition or SVD. So any function on $[0,\infty)^n$ that is a good substitute for "count the number of non-zero entries" should give, when applied to the diagonalization of $A*A$, give some substitute for rank which might fit your purposes $\endgroup$
    – Yemon Choi
    Commented Jun 16, 2016 at 22:22

1 Answer 1

Reset to default
4
$\begingroup$

The notion of stable rank is often used in the low-rank matrix approximation literature to offer a more tractable surrogate for rank. This does not necessarily satisfy all your requirements, but might be still suitable for your purposes.

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .