0
$\begingroup$

Let $H$ be a Hilbert space and let $a,b\in B(H)$ be such that $${\rm Tr}(l(a))={\rm Tr}(r(a))<\infty , {\rm Tr}(l(b))={\rm Tr}(r(b))<\infty,$$ $${\rm Tr}(l(a)+l(b)) ={\rm Tr}(l(\alpha_1 a +\alpha_2 b ))$$ for any $0\ne \alpha_1,\alpha_2\in \mathbb{R}$, where ${\rm Tr}$ denotes the standard trace on $B(H)$ and $l(\cdot),r(\cdot)$ denote the left and the right-support respectively. Can we say that $l(a)l(b)=r(a)r(b)=0$? If we replace $B(H)$ with $\ell_\infty$, then the above equality is clearly true. However, I am not sure about the case for $B(H)$.

$\endgroup$
4
  • 1
    $\begingroup$ If you require $\text{Tr}(l(a)+l(b)) = \text{Tr}(l(\alpha_1 a + \alpha_2 b))$ for really all $\alpha_1,\alpha_2$, then also for $\alpha_1 = \alpha_2 = 0$, which forces $a=b=0$. You probably want to modify the question? $\endgroup$ Commented Apr 14, 2023 at 8:55
  • $\begingroup$ @StefaanVaes THX Stefaan, you are right about that. $\endgroup$
    – user92646
    Commented Apr 14, 2023 at 9:52
  • $\begingroup$ It is best not to include displayed math inside a title. $\endgroup$ Commented Apr 14, 2023 at 10:16
  • $\begingroup$ @GeraldEdgar I see! THX $\endgroup$
    – user92646
    Commented Apr 14, 2023 at 10:17

1 Answer 1

3
$\begingroup$

This already fails for $2 \times 2$ matrices. Take the rank one matrices $a = \begin{pmatrix} 1 & 0 \\ 0 & 0 \end{pmatrix}$ and $b = \begin{pmatrix} 1 & 1 \\ 1 & 1 \end{pmatrix}$. For all $\alpha \neq 0$ and $\beta \neq 0$, the matrix $\alpha a + \beta b$ is invertible, so that $\text{Tr}(l(\alpha a + \beta b)) = 2$. Also $\text{Tr}(l(a) + l(b)) = 2$. But the ranges of $a$ and $b$ are not orthogonal.

$\endgroup$
1
  • $\begingroup$ Thx very much! the situation in the noncommutative setting is so different with the commutative setting. $\endgroup$
    – user92646
    Commented Apr 14, 2023 at 10:14

You must log in to answer this question.

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