2
$\begingroup$

I have the Hamiltonian evolution operator for a two qubit system :

\begin{equation} \hat{U}= e^{-i\gamma H}\;,\quad H = \frac{1}{2}(I - Z\otimes Z) \end{equation} where the $Z$ gate is applied on both qubits, ($I=I\otimes I$).

I would like to express this evolution operator in terms of CNOT gate and rotations.

I was told that $e^{i \frac{\gamma}{2}Z\otimes Z}$ could be decomposed as follows : \begin{equation} e^{i \frac{\gamma}{2}Z\otimes Z}=\text{CX}\;(I\otimes R_z(\gamma))\;\text{CX}\;,\quad R_z(\gamma) = e^{-i\frac{\gamma}{2}Z} \end{equation} where the CNOT gate operates on the second qubit each time (first qubit being the control one) and the rotation applies to the second one.

I found a related question here but I wonder if this can be simply demonstrated with the CNOT gate decomposition : \begin{equation} \text{CX}= e^{i\frac{\pi}{4}(I -Z)\otimes (I-X)} \end{equation}

The development is the following : \begin{equation} \begin{split} \text{CX}\;(I\otimes e^{-i\frac{\gamma}{2}Z})\;\text{CX} &= e^{i\frac{\pi}{4}(I -Z)\otimes (I-X)}\;(I\otimes e^{-i\frac{\gamma}{2}Z})\; e^{i\frac{\pi}{4}(I -Z)\otimes (I-X)}\\ &=e^{i\frac{\pi}{4}(I -Z)\otimes (I-X)}\;e^{-i\frac{\gamma}{2}I\otimes Z}\; e^{i\frac{\pi}{4}(I -Z)\otimes (I-X)} \end{split} \end{equation}

Then, I am not sure how to proceed, it seems to me not trivial to end up with $e^{i \frac{\gamma}{2}Z\otimes Z}$. This is the question. If the matrices within exponential do not commute, I don't think I can go further.

$\endgroup$
2
  • $\begingroup$ You seem to have already answered your own question, so what are you actually asking? $\endgroup$
    – DaftWullie
    Commented Feb 23, 2023 at 12:09
  • $\begingroup$ I tried to demonstrate the second formula by expanding CX gates with its decomposition in Pauli basis, but i was neither sure of the developments nor of if it is really achievable in this way. The related question I mentioned takes another way by using another expression of the CNOT gate. $\endgroup$
    – deb2014
    Commented Feb 23, 2023 at 12:58

1 Answer 1

1
$\begingroup$

To verify the action of this circuit, I suggest computing how it acts on each of the 4 basis states $|00\rangle,|01\rangle,|10\rangle$ and $|11\rangle$.

For example $$ |00\rangle\xrightarrow{CX}|00\rangle\xrightarrow{R_Z}e^{-i\gamma/2}|00\rangle\xrightarrow{CX}e^{-i\gamma/2}|00\rangle. $$ Work this out for all 4 basis states. Is this consistent with what your gate is supposed to do?


You've asked for an algebraic method. While I don't entirely understand what is sought, here is another method...

Let $P_0=|0\rangle\langle 0|$ and $P_1=|1\rangle\langle 1|$. Then $$ CX=P_0\otimes I+P_1\otimes X $$ and $$ I\otimes R_Z=(P_0+P_1)\otimes R_Z. $$ It is important to recognise that $P_0P_1=0$. Hence, when I multiply together the 3 circuit elements, I get $$ P_0\otimes R_Z+P_1\otimes(XR_ZX)=P_0\otimes R_Z+P_1\otimes(R_Z^\dagger)=\frac12(I+Z)\otimes R_Z+\frac12(I-Z)\otimes R_Z^\dagger $$ Regroup the terms, $$ I\otimes(R_Z+R_Z^\dagger)+Z\otimes (R_Z-R_Z^\dagger)=I\otimes I\cos(\gamma/2)-i\sin(\gamma/2)Z\otimes Z=e^{-i\gamma Z\otimes Z/2} $$ but this all seems rather painful!

$\endgroup$
2
  • $\begingroup$ Yes, this is the way I was told the equality. My question was : is there an algebraic way to demonstrate this equality. $\endgroup$
    – deb2014
    Commented Feb 23, 2023 at 14:27
  • $\begingroup$ But what are the bounds on what you mean by that? Why not simply write out the matrices and multiply them? $\endgroup$
    – DaftWullie
    Commented Feb 23, 2023 at 14:38

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.