I'm trying to understand the steps in a certain proof, where I think the following claim is used: $S( |i \rangle \langle i| \otimes \rho) = S(\rho)$, where $S()$ is the entropy, $|i \rangle$ is a state from some orthonormal basis, and $\rho$ is a density operator. Here's my attempt to see why this holds:

$$ \begin{align} S( |i \rangle \langle i| \otimes \rho) &= -tr\left[\left(|i \rangle \langle i| \otimes \rho\right) \times \log \left(|i \rangle \langle i| \otimes \rho\right)\right]\\ &= -tr[(|i \rangle \langle i| \times \log |i \rangle \langle i|) \otimes \rho \log \rho]\\ &= -tr[(|i \rangle \langle i| \times \log |i \rangle \langle i|)] \times tr[\rho \log \rho]\\ &= 0 \times S(\rho) \end{align}$$

so I am clearly making a mistake somewhere, that $0$ should not be there. The first equality comes from the definition of $S()$, the second because $(A \otimes B)(C \otimes D) = AC \otimes BD$, the third because $tr(A \otimes B) = tr(A) \times tr(B)$. Are any of these incorrect?

  • 1
    $\begingroup$ You also put $log (A \otimes B) = log A \otimes log B$ which got fed in as $C$ and $D$ in the second. $\endgroup$
    – AHusain
    Commented Jan 17, 2017 at 0:45

1 Answer 1


$$\log(A\otimes B)=\log A \otimes I + I\otimes \log B$$ You can show this by diagonalizing $A$ and $B$ and using the idea that the matrix log of a diagonal matrix is the log of the elements. In components,

$$\log(A\otimes B)_{ii',jj'}=\log(A_i B_{i'})I_{ij}I_{i'j'}=(\log A_i I_{ij})I_{i'j'}+I_{ij}(\log B_{i'}I_{i'j'})$$

Then if you use your correct rules for tensor product manipulation in your question you will prove the statement.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.