Given two distinct and noninteracting quantum mechanical systems $\mathfrak{S}\_1$ and $\mathfrak{S}\_2$ with state spaces $\mathcal H\_1$ and $\mathcal H\_2$, respectively, the state space of the combined system $\mathcal S\_1+\mathcal S\_2$ is the tensor product Hilbert space $\mathcal H=\mathcal H\_1\otimes\mathcal H\_2$. Density operators $W\in\mathcal D(\mathcal H)$, and effects $F\in\mathcal E(\mathcal H)$. Similarly, there are corresponding symbols $W\_i\in\mathcal D(\mathcal H\_i), F\_i\in\mathcal E(\mathcal H\_i)$ for subsystems $\mathfrak{S}\_i(i=1,2)$, respectively.

Given any quantum operation, $\Phi: \mathcal D(\mathcal H)\rightarrow \mathcal D(\mathcal H)$, of the composite system $\mathcal S\_1+\mathcal S\_2$.

Problem: (1) Do there exist whether or not two quantum operation $\phi\_1$ and $\phi\_2$, of the subsystems $\mathfrak{S}\_1$ and $\mathfrak{S}\_2$, respectively, such that the following diagram is commutative:

$$ \begin{diagram} \node{\mathcal D(\mathcal H\_1)} \arrow[4]{e,t}{\phi\_1}\node[4]{\mathcal D(\mathcal H\_1)}\\ \node{}\\ \node{\mathcal D(\mathcal H\_1\otimes\mathcal H\_2)} \arrow[2]{n,l}{Tr\_2} \arrow[4]{e,t}{\Phi} \arrow[2]{s,l}{Tr\_1} \node[4]{\mathcal D(\mathcal H\_1\otimes\mathcal H\_2)} \arrow[2]{s,r}{Tr\_1} \arrow[2]{n,r}{Tr\_2} \\ \node{}\\ \node{\mathcal D(\mathcal H\_2)} \arrow[4]{e,b}{\phi\_2} \node[4]{\mathcal D(\mathcal H\_2)} \end{diagram} $$ i.e. $$\begin{eqnarray} Tr\_2(\Phi(W))&=&\frac{tr(\Phi(W))}{tr(\phi\_1(Tr\_2(W)))}\phi\_1(Tr\_2(W)),\\ Tr\_1(\Phi(W))&=&\frac{tr(\Phi(W))}{tr(\phi\_2(Tr\_1(W)))}\phi\_2(Tr\_1(W)), \end{eqnarray} $$ where $\phi\_i: \mathcal D(\mathcal H\_i)\rightarrow \mathcal D(\mathcal H\_i)(i=1,2)$ and $Tr\_i: \mathcal D(\mathcal H)\rightarrow \mathcal D(\mathcal H\_i)$ is a partial trace with respect to the subsystem $\mathfrak{S}\_i(i=1,2)$.

(2) If quantum operation $\phi_1$ and $\phi_2$ exist, give the relationship among the quantum operations $\Phi, \phi_{1}$ and $\phi_2$.

  • $\begingroup$ I have edited the slab of LaTeX which you seemed to have pasted in without using the preview function. Please have a look at the guidelines as to what MO does and doesn't support, and how to get round them. $\endgroup$
    – Yemon Choi
    Feb 11, 2010 at 10:17
  • $\begingroup$ @Yemon: Your (heroic!) edit doesn't seem too successful to me. I am getting this text: Unknown environment "diagram". $\endgroup$ Feb 11, 2010 at 13:58
  • $\begingroup$ @Harald: if you saw what happened before my edit, I think the present version is a considerable improvement:) I had insufficient zeal/charity for trying to work out what the diagram was supposed to depict and then recoding it $\endgroup$
    – Yemon Choi
    Feb 11, 2010 at 19:56
  • $\begingroup$ @Yemon: Yes, I had looked at the edit history. That's why I said “heroic”. Not complaining, really. $\endgroup$ Feb 12, 2010 at 12:34

1 Answer 1


If I understand the question right, essentially what is being asked is: for a general quantum operation on a system with two subsystems, can you learn what happens to subsystem 1 just by looking at the initial state of subsystem 1, and what happens to subsystem 2 just by looking at the initial state of subsystem 2. If this is what is being asked, the answer is No: consider the case where the two subsystems have the same dimension and the quantum operation $\Phi$ which swaps the subsystems.


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.