10
$\begingroup$

While trying to simplify a certain 2-qubit quantum circuit, I managed to get it down to this:

But by inspecting the corresponding two-qubit unitary directly, I can come up with the arguably simpler:

where the rotation operator "moved" to the first qubit. I am using the convention $ R_{\theta} := R_y(2\theta) = \begin{pmatrix} \cos\theta & -\sin\theta \\ \sin\theta & \cos\theta \end{pmatrix} $. The rotation $R_{\pi/4}$ is thus $=XH$.

How can I show that the two circuits are equivalent by "elementary" circuit identities, instead of just verifying that they amount to the same unitary? I tried many different simplifications without success.

$\endgroup$

3 Answers 3

5
$\begingroup$

Why not make it even smaller?

One two qubit

Initial state:

enter image description here

Change rotation basis:

enter image description here

Phase the parity on the top bit instead of the bottom bit:

enter image description here

Apply XZ = iY to the adjacent CX CZ operations:

enter image description here

Move the middle CNOT leftward, changing its interaction basis from XZ to XY to ZY to ZY as it moves through the single qubit gates, resulting in it become a CY:

enter image description here

Cancel the adjacent CYs, then propagate the non-parameterized rotations rightward. This changes the basis of the rotation back to Y, and transforms the NOTC into a CY.

enter image description here

Cancel the two X rotations and you're done.

$\endgroup$
5
$\begingroup$

I am going to assume some usual transformations such as the involution of $H$ and CNOT. Rewriting the two gates $R$ with more standard gates, a possible derivation goes like this (details for equalities (1), (2) and (3) are given below):

enter image description here

I used in particular the identity:

enter image description here

from which I also derived:

enter image description here

I also used:

enter image description here

A very powerful tool for doing this kind of graphical proof is the ZX-Calculus (https://arxiv.org/abs/0906.4725), which I actually used for reasoning on your problem here, before turning the answer in circuit form.

$\endgroup$
3
$\begingroup$

If $U_1 = U_2$, then $U_1 U_2^{\dagger} = I$. So let's add to the first circuit the inverse/dagger of the second one:

This whole thing should be an identity. Note that, for this, we should prove that whole circuit except last $R_y(-\theta) I$ should be equal to $R_y(\theta) I$. By remembering that the SWAP gate consists of 3 CNOT gates and $\text{SWAP} \cdot I R_y(\theta) \cdot\text{SWAP} = R_y(\theta) I$, we will have:

This should be equal to $R_y(\theta) I$. This circuit is equal to $U \cdot R_y(\theta) I \cdot U^\dagger$, where $U$ is this circuit:

Now we should try to prove that $U \cdot R_y(\theta) I \cdot U^\dagger = R_y(\theta) I$. After simplifying the $U$ I have obtained this circuit:

where I took into account that $HXH = Z$ and $I \otimes H \cdot \text{CNOT} \cdot I \otimes H = CZ$ (note that I have omitted the $\otimes$ sign in the previous expressions). If my calculations are right, then $U = \frac{1}{\sqrt{2}}(I \otimes I+iY \otimes X)$. So, by omitting again the $\otimes$ sign:

$$U \cdot R_y(\theta) I \cdot U^\dagger = \frac{1}{2}\left[II+iY X\right]\left[\cos(\theta)I I - i\sin(\theta)Y I\right]\left[I I-iY X\right] = R_y(\theta) I$$

So, the circuits are equivalent.

$\endgroup$

Your Answer

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

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